Browse the corpus

Breast cancer is the most common cancer in young adult women (age, ≤40 years) in the UK, with over 2000 new cases annually.1 Young women have been found to develop more aggressive tumors coupled with lower survival and higher local-recurrence rates (LRR) than older women,2–9 and this may be a particular issue in the developing world where a greater proportion of breast cancer appears in women of young age.7,8 The choice between mastectomy and breast-conserving surgery (BCS) in young women is not often a straightforward decision for clinician and/or patient.10 BCS is associated with better quality of life but higher LRR,4,11 although a meta-analysis of mostly registry and database studies in patients <40 years suggests equivalent disease-free and overall survival,12 whereas very young age (<35 years) has been considered a relative contraindication to BCS.13 Although randomized controlled trials (RCTs) suggest equivalent survival for mastectomy and BCS, very few young patients were included in these analyses.6,14 Indeed, young women are not routinely analyzed or reported separately in individual RCTs of BCS versus mastectomy, and any that do have very few women ≤40 years presented.12

ndomized controlled trials (RCTs) suggest equivalent survival for mastectomy and BCS, very few young patients were included in these analyses.6,14 Indeed, young women are not routinely analyzed or reported separately in individual RCTs of BCS versus mastectomy, and any that do have very few women ≤40 years presented.12 Emerging evidence suggests a possible survival advantage for mastectomy in BRCA gene-mutation carriers.15 Young patients are more likely to be BRCA-mutation positive5 and retrospective cohort studies suggest that LRR are higher for BCS compared with mastectomy.6,10,11,16 Although family history does not affect clinical outcome in young patients, it appears to affect surgical type selection, and it is unknown if family history of breast cancer will influence local recurrence.17

tive5 and retrospective cohort studies suggest that LRR are higher for BCS compared with mastectomy.6,10,11,16 Although family history does not affect clinical outcome in young patients, it appears to affect surgical type selection, and it is unknown if family history of breast cancer will influence local recurrence.17 The effect of radiotherapy plays a key role in the treatment of younger breast cancer patients. The Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) found that women <40 years undergoing BCS had the highest incidence of recurrence and the largest benefit from radiotherapy; with the 10-year recurrence rate (local or distant) significantly lower compared with those without radiotherapy (36.1% vs 60.7%, respectively, P = 0.00009).18 Similar results were seen in an RCT investigating the benefit of radiotherapy boost after BCS, where the largest absolute benefit was seen in patients ≤40 years with a significant relative-risk reduction for boost (P = 0.003).19 Likewise, an RCT of premenopausal women undergoing mastectomy with/without radiotherapy showed that irradiation after mastectomy significantly improved outcomes, even after controlling for clinical and pathological factors,20 and a Canadian population registry study of 588 women <35 years found that LRR were reduced for postmastectomy radiation.21

usal women undergoing mastectomy with/without radiotherapy showed that irradiation after mastectomy significantly improved outcomes, even after controlling for clinical and pathological factors,20 and a Canadian population registry study of 588 women <35 years found that LRR were reduced for postmastectomy radiation.21 These differences observed in the effect of radiotherapy, and the trend towards young patients having bilateral mastectomy as part of their initial cancer treatment, demonstrate that an important question remains about surgical type and outcomes in this age group. Moreover, local recurrence is very important in young breast cancer patients as there are a fewer competing risks and, other than their breast cancer, their life expectation is longer. There are no large prospective cohort studies reporting local recurrence in this age group, and a dedicated RCT comparing mastectomy versus BCS in young women is unlikely. A large prospective cohort study may offer the best level of evidence, minimizing inclusion bias, to guide management, and enable a comparison of local recurrence with disease-free survival. The Prospective study of Outcomes in Sporadic versus Hereditary breast cancer (POSH) is an observational cohort of 3000 young women with breast cancer, and is representative of the UK breast cancer population.22 We have not previously reported local-recurrence outcomes, and the aim of this analysis was therefore to report breast ipsilateral LRR in the POSH study to determine whether acceptable rates are found in a large cohort of young patients, and what factors, including surgical type, affect these outcomes in this age group.

e have not previously reported local-recurrence outcomes, and the aim of this analysis was therefore to report breast ipsilateral LRR in the POSH study to determine whether acceptable rates are found in a large cohort of young patients, and what factors, including surgical type, affect these outcomes in this age group. METHODS Study Population POSH (MREC: 00/06/69) is a multicenter prospective observational cohort study of 3000 young women diagnosed with breast cancer in the UK between 2000 and 2008 (http://www.southampton.ac.uk/medicine/research/posh.page). All patients received treatment according to local protocols. The detailed study protocol was published in 2007,22 and the cohort previously described.23 For this analysis, type of surgery was defined as the final oncological surgery to the breast for example, if a patient had BCS followed by mastectomy ≤3 months; this was classed as a mastectomy. A mastectomy performed >3 months after primary treatment in the absence of local disease-recurrence was considered risk reducing rather than oncological. Analyses of risk-reducing surgery will be the subject of future work. Margin status was the final surgical margin after oncological operation(s), and a positive margin was defined according to American Society for Clinical Oncology (ASCO) guidance as tumor at the margin (ie, tumor on ink).24 This article presents analyses conducted on follow-up data from the POSH cohort received until June 26, 2015.

s was the final surgical margin after oncological operation(s), and a positive margin was defined according to American Society for Clinical Oncology (ASCO) guidance as tumor at the margin (ie, tumor on ink).24 This article presents analyses conducted on follow-up data from the POSH cohort received until June 26, 2015. Statistical Analysis All analyses were conducted according to a prespecified plan in line with published guidance.25 Patients with metastatic disease at presentation were excluded. Summary statistics were used to describe the cohort and key characteristics were compared by surgical type using Pearson χ2 tests or Mann-Whitney U tests. All reported P-values were 2-sided.

cording to a prespecified plan in line with published guidance.25 Patients with metastatic disease at presentation were excluded. Summary statistics were used to describe the cohort and key characteristics were compared by surgical type using Pearson χ2 tests or Mann-Whitney U tests. All reported P-values were 2-sided. Study endpoints were inbreast ipsilateral local-recurrence interval (LRI), distant disease-free interval (DDFI), and overall survival (OS). LRI was defined as time from date of diagnosis to date of local recurrence (either an ipsilateral recurrence or ipsilateral new primary, whichever event occurred first after BCS or chest-wall recurrence after mastectomy). The local-recurrence event was counted as an event if the date of the nonevent (death from breast cancer, distant metastases, ipsilateral local axillary recurrence, ipsilateral regional nodes recurrence, and/or contralateral recurrence, if/where applicable) was >3 months after the date of the local-recurrence event. If the date of the nonevent was ≤3 months after the local recurrence event then the patient was censored at the date of nonevent. Deaths from other cancers after local recurrence did not affect the event. DFFI was defined as time from breast cancer diagnosis to distant metastases or death from breast cancer; deaths from other causes were censored at the time of death. OS was defined as time from breast cancer diagnosis to death from any cause.

t. Deaths from other cancers after local recurrence did not affect the event. DFFI was defined as time from breast cancer diagnosis to distant metastases or death from breast cancer; deaths from other causes were censored at the time of death. OS was defined as time from breast cancer diagnosis to death from any cause. Nelson-Aalen cumulative-hazard plots were used to describe LRI and Kaplan-Meier plots were used to describe DDFI and OS. Univariable analyses (UVA) and multivariable analyses (MVA) were carried out using Cox proportional-hazards models, or Flexible Parametric Survival Models (FPSMs) for models which involved time-varying hazards.26 Covariates included in the MVA models included age at diagnosis (fitted as a continuous variable), tumor size, focality, nodal (N) stage, histological grade, ER and HER2 tumor status, adjuvant radiotherapy, adjuvant hormone therapy, and surgical margins, regardless of significance. Patients treated with neoadjuvant chemotherapy were included in UVA but excluded from all MVA because of difficulties in classifying pathological T and N staging for these patients. For each FPSMs, we explored varying degrees of freedom for the baseline-hazard rate and time-dependent effect to obtain the best model fit. All analyses were performed using STATA v13.1 (StataCorp, College Station, TX, USA) on records with complete data (levels of missing data were reported).

Nelson-Aalen cumulative-hazard plots were used to describe LRI and Kaplan-Meier plots were used to describe DDFI and OS. Univariable analyses (UVA) and multivariable analyses (MVA) were carried out using Cox proportional-hazards models, or Flexible Parametric Survival Models (FPSMs) for models which involved time-varying hazards.26 Covariates included in the MVA models included age at diagnosis (fitted as a continuous variable), tumor size, focality, nodal (N) stage, histological grade, ER and HER2 tumor status, adjuvant radiotherapy, adjuvant hormone therapy, and surgical margins, regardless of significance. Patients treated with neoadjuvant chemotherapy were included in UVA but excluded from all MVA because of difficulties in classifying pathological T and N staging for these patients. For each FPSMs, we explored varying degrees of freedom for the baseline-hazard rate and time-dependent effect to obtain the best model fit. All analyses were performed using STATA v13.1 (StataCorp, College Station, TX, USA) on records with complete data (levels of missing data were reported). RESULTS Patient Characteristics and Definitive Surgery Information The POSH study recruited 3095 patients across the United Kingdom, and of 2882 included in this analysis (Fig. 1), 1464 (50.8%) underwent mastectomy and 1395 (48.4%) BCS. All patients included underwent surgery to the axilla (axillary dissection, sentinel node biopsy, or sample ± axillary dissection). Twenty-three (0.8%) patients underwent lymph node surgery only, with no surgery to the breast. Table 1 shows baseline demographics by surgical type. Median age at diagnosis was 36 years for mastectomy and BCS. Family history of breast cancer was reported significantly more for mastectomy compared with BCS (52.1% vs 48.1%, P = 0.037), and surveillance-detected tumors were more frequent for mastectomy than BCS (1.5% vs 0.6%). However, no significant differences were observed between surgical type for BMI and ethnicity.

nd BCS. Family history of breast cancer was reported significantly more for mastectomy compared with BCS (52.1% vs 48.1%, P = 0.037), and surveillance-detected tumors were more frequent for mastectomy than BCS (1.5% vs 0.6%). However, no significant differences were observed between surgical type for BMI and ethnicity. FIGURE 1 Flow chart for POSH study; local-recurrence analyses. TABLE 1 Baseline Demographic Information for All Patients by Surgery Type

nd BCS. Family history of breast cancer was reported significantly more for mastectomy compared with BCS (52.1% vs 48.1%, P = 0.037), and surveillance-detected tumors were more frequent for mastectomy than BCS (1.5% vs 0.6%). However, no significant differences were observed between surgical type for BMI and ethnicity. FIGURE 1 Flow chart for POSH study; local-recurrence analyses. TABLE 1 Baseline Demographic Information for All Patients by Surgery Type Characteristic Mastectomy Breast Conserving Surgery Total* P† (n = 1464) (n = 1395) (n = 2882) Age at diagnosis, y 0.868 Median 36 36 36 Range 18 to 40 19 to 40 18 to 40 IQR 33 to 38 34 to 38 33 to 38 Missing 0 0 0 Body mass index 0.154 Median 24.5 24.8 24.6 Range 16.5 to 59.5, 16.8 to 55.9, 16.5 to 59.5, IQR 22.0 to 28.1 22.1 to 28.4 22.1 to 28.4 Missing 44 (3.0%) 64 (4.6%) 108 (3.7%) Race/ethnicity 0.436 White 1324 (91.9%) 1284 (93.2%) 2625 (92.4%) Black 63 (4.4%) 44 (3.2%) 112 (3.9%) Asian 43 (3.0%) 41 (3.0%) 84 (3.0%) Other 10 (0.7%) 9 (0.7%) 20 (0.7%) Missing 24 (1.6%) 17 (1.2%) 41 (1.4%) Family history 0.037 No 672 (47.9%) 690 (51.9%) 1378 (50.0%) Yes 731 (52.1%) 640 (48.1%) 1376 (50.0%) Missing 61 (4.2%) 65 (4.7%) 128 (4.4%) Presentation <0.001 Symptomatic 1424 (97.7%) 1380 (99.4%) 2826 (98.5%) Screen detected 22 (1.5%) 8 (0.6%) 30 (1.0%) Other 12 (0.8%) 0 12 (0.4%) Missing 6 (0.4%) 7 (0.5%) 14 (0.5%) Histological grade 0.005 Grade 1 68 (4.8%) 93 (6.8%) 161 (5.7%) Grade 2 515 (36.2%) 429 (31.3%) 948 (33.7%) Grade 3 840 (59.0%) 848 (61.9%) 1703 (60.6%) Missing 41 (2.8%) 25 (1.8%) 70 (2.4%) Histological type <0.001 Ductal 1246 (86.3%) 1230 (89.1%) 2494 (87.7%) Lobular 85 (5.9%) 44 (3.2%) 131 (4.6%) Ductal and lobular 50 (3.5%) 24 (1.7%) 74 (2.6%) Other 83 (5.7%) 97 (7.0%) 183 (6.4%) Missing 20 (1.4%) 15 (1.1%) 37 (1.3%) Surgical margin <0.001 0 98 (8.9%) 113 (10.0%) 211 (9.4%) ≥0 to <1 97 (8.8%) 126 (11.1%) 223 (10.0%) ≥1 to ≤5 438 (39.6%) 624 (55.0%) 1062 (47.4%) >5 472 (42.7%) 272 (24.0%) 745 (33.2%) Missing 359 (24.5%) 260 (18.6%) 641 (22.2%) EIC‡ <0.001 Negative 1010 (72.8%) 1178 (86.8%) 2189 (79.7%) Positive 378 (27.2%) 179 (13.2%) 557 (20.3%) Missing 76 (5.2%) 38 (2.7%) 136 (4.7%) Lymphovascular invasion <0.001 Absent 614 (45.1%) 784 (59.9%) 1402 (52.4%) Present 747 (54.9%) 524 (40.1%) 1276 (47.6%) Missing 103 (7.0%) 87 (6.2%) 204 (7.1%) Number of positive nodes <0.001 0 549 (37.6%) 837 (60.5%) 1389 (48.7%) 1–3 532 (36.5%) 404 (29.2%) 940 (33.0%) 4–9 246 (16.9%) 99 (7.2%) 346 (12.1%) 10+ 132 (9.0%) 43 (3.1%) 175 (6.1%) Missing 5 (0.3%) 12 (0.9%) 32 (1.1%) ER status <0.001 Negative 449 (30.7%) 516 (37.2%) 975 (34.0%) Positive 1013 (69.3%) 870 (62.8%) 1

Number of positive nodes <0.001 0 549 (37.6%) 837 (60.5%) 1389 (48.7%) 1–3 532 (36.5%) 404 (29.2%) 940 (33.0%) 4–9 246 (16.9%) 99 (7.2%) 346 (12.1%) 10+ 132 (9.0%) 43 (3.1%) 175 (6.1%) Missing 5 (0.3%) 12 (0.9%) 32 (1.1%) ER status <0.001 Negative 449 (30.7%) 516 (37.2%) 975 (34.0%) Positive 1013 (69.3%) 870 (62.8%) 1 896 (66.0%) Missing 2 (0.1%) 9 (0.6%) 11 (0.4%) PR status 0.009 Negative 478 (40.7%) 519 (46.1%) 1008 (43.5%) Positive 697 (59.3%) 607 (53.9%) 1309 (56.5%) Missing 289 (19.7%) 269 (19.3%) 565 (19.6%) HER2 status <0.001 Negative 906 (69.6%) 918 (75.9%) 1839 (72.7%) Positive 395 (30.4%) 292 (24.1%) 691 (27.3%) Missing 163 (11.1%) 185 (13.3%) 352 (12.2%) Focality <0.001 Localised 707 (53.5%) 1136 (87.5%) 1845 (70.3%) Multifocal 615 (46.5%) 163 (12.5%) 779 (29.7%) Missing 142 (9.7%) 96 (6.9%) 258 (9.0%) Maximum invasive tumor size (mm) <0.001 Median 28.5 19.0 22.0 Range 0.0 to 199.0, 0.0 to 90.0, 0.0 to 199.0, IQR 19.0 to 43.0 14.0 to 25.0 15.0 to 33.0 Missing 93 (6.4%) 45 (3.2%) 160 (5.6%) Maximum overall (invasive + in situ) tumor size, mm <0.001 Median 37.0 20.5 27.0 Range 0.0 to 199.0, 0.0 to 115.0, 0.0 to 199.0, IQR 25.0 to 55.0 15.0 to 27.0 18.0 to 40.0 Missing 68 (4.6%) 38 (2.7%) 128 (4.4%) Number of operations, categorical <0.001 1 1047 (71.5%) 1105 (79.2%) 2174 (75.4%) 2 361 (24.7%) 279 (20.0%) 641 (22.2%) 3 53 (3.6%) 11 (0.8%) 64 (2.2%) 4 2 (0.1%) 0 2 (0.1%) 5 1 (0.1%) 0 1 (0.0%) Missing 0 0 0 Chemotherapy treatment period <0.001 Adjuvant 1088 (74.3%) 1055 (75.6%) 2149 (74.6%) Neoadjuvant 275 (18.8%) 155 (11.1%) 447 (15.5%) Palliative 1 (0.1%) 0 1 (0.0%) Not applicable 100 (6.8%) 185 (13.3%) 285 (9.9%) Missing 0 0 0 Adjuvant trastazumab N/A No/missing 1255 (85.7%) 1246 (89.3%) 2523 (87.5%) Yes 209 (14.3%) 149 (10.7%) 359 (12.5%) Missing 0 0 0 Adjuvant radiotherapy N/A No/missing 458 (31.3%) 56 (4.0%) 525 (18.2%) Yes 1006 (68.7%) 1339 (96.0%) 2357 (81.8%) Missing 0 0 0 Adjuvant hormone treatment N/A No/missing 490 (33.5%) 556 (39.9%) 1059 (36.7%) Yes 974 (66.5%) 839 (60.1%) 1823 (63.3%) Missing 0 0 0 EIC indicates extensive intraductal component; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; IQR, interquartile range; PR, progesterone receptor; U/H, underweight/healthy.

ormone treatment N/A No/missing 490 (33.5%) 556 (39.9%) 1059 (36.7%) Yes 974 (66.5%) 839 (60.1%) 1823 (63.3%) Missing 0 0 0 EIC indicates extensive intraductal component; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; IQR, interquartile range; PR, progesterone receptor; U/H, underweight/healthy. *Total column includes data from the whole cohort that is, BCS, mastectomy, and other (23 patients). †P-value obtained using Pearson χ2 test (for categorical variables) or Mann-Whitney test (for continuous variables). ‡EIC defined as positive where the total tumor in-situ size is ≥25% the size of the total tumor size (or where the total tumor invasive size is <75% the size of the total tumor size). Tumor Pathology Significant differences in grade and focality were found between mastectomy and BCS (P = 0.005 and P < 0.001, respectively). Patients undergoing mastectomy had larger tumors were more likely to be human epidermal growth factor receptor 2+ (HER2+) and with a higher proportion of Extensive Intraductal Component positive (EIC+) compared with BCS (P < 0.001 in all cases). Patients undergoing mastectomy had a significantly higher proportion of ER+ and/or PR+ tumors than BCS (estrogen receptor, ER: 69.3% vs 62.8%, P < 0.001; progesterone receptor, PR: 59.3% vs 53.9%, P = 0.009, respectively).

Tumor Pathology Significant differences in grade and focality were found between mastectomy and BCS (P = 0.005 and P < 0.001, respectively). Patients undergoing mastectomy had larger tumors were more likely to be human epidermal growth factor receptor 2+ (HER2+) and with a higher proportion of Extensive Intraductal Component positive (EIC+) compared with BCS (P < 0.001 in all cases). Patients undergoing mastectomy had a significantly higher proportion of ER+ and/or PR+ tumors than BCS (estrogen receptor, ER: 69.3% vs 62.8%, P < 0.001; progesterone receptor, PR: 59.3% vs 53.9%, P = 0.009, respectively). Treatment and Surgery Information Patients undergoing mastectomy had a higher frequency of negative margins compared with BCS (P < 0.001). Specifically, the proportion of margins >5 mm was shown to be higher (42.7% vs 24.0%, respectively), whereas the proportion of margins 1 to 5 mm was lower for BCS (39.6% vs 55.0%, respectively).

Surgery Information Patients undergoing mastectomy had a higher frequency of negative margins compared with BCS (P < 0.001). Specifically, the proportion of margins >5 mm was shown to be higher (42.7% vs 24.0%, respectively), whereas the proportion of margins 1 to 5 mm was lower for BCS (39.6% vs 55.0%, respectively). Although the median number of operations was one for mastectomy and BCS, the distribution was significantly different (P < 0.001), with a higher proportion of patients undergoing mastectomy having more than one surgery (28.5% vs 20.8%, P < 0.001). Only 11.1% of BCS patients underwent neoadjuvant chemotherapy compared with 18.8% for mastectomy. Adjuvant radiotherapy was given to 68.7% of patients undergoing a mastectomy. In 56 patients undergoing BCS, we were unable to confirm adjuvant breast radiotherapy. We cannot exclude that these patients had radiotherapy to the breast at a different center and that the information was not recorded, nor can we confirm that these patients did not receive adjuvant breast radiotherapy. However, these patients were analyzed as not having radiotherapy. Nine-hundred (61.5%) patients undergoing mastectomy had chest-wall radiotherapy (CWR-XRT) whereas 977 (70.0%) patients undergoing BCS had a radiotherapy boost. In patients undergoing BCS, no clear association was seen between margin status (>0/negative vs 0/positive) and provision of a radiotherapy boost (data not shown).

e-hundred (61.5%) patients undergoing mastectomy had chest-wall radiotherapy (CWR-XRT) whereas 977 (70.0%) patients undergoing BCS had a radiotherapy boost. In patients undergoing BCS, no clear association was seen between margin status (>0/negative vs 0/positive) and provision of a radiotherapy boost (data not shown). Missing data were similar across surgical types and low for most demographic information. Exceptions were PR, HER2, and surgical margin information, with up to 19.7%, 13.3%, and 24.5% missing, respectively.

e-hundred (61.5%) patients undergoing mastectomy had chest-wall radiotherapy (CWR-XRT) whereas 977 (70.0%) patients undergoing BCS had a radiotherapy boost. In patients undergoing BCS, no clear association was seen between margin status (>0/negative vs 0/positive) and provision of a radiotherapy boost (data not shown). Missing data were similar across surgical types and low for most demographic information. Exceptions were PR, HER2, and surgical margin information, with up to 19.7%, 13.3%, and 24.5% missing, respectively. Follow Up and Survival Median follow up was 7.3 years for mastectomy, BCS, and overall. There were 139 local-recurrence events compared with 752 DDFI events overall, demonstrating that the majority of events experienced were because of distant metastases or death from breast cancer (Fig. 1). Ninety-five local-recurrence events were experienced for patients undergoing BCS (6.8% of these patients), compared with just 40 (2.7%) for mastectomy. Two-hundred and sixty DDFI events were experienced for BCS (18.6%), and 485 (33.1%) for mastectomy. Similar numbers were found for OS, with 232 (16.6%) and 431 (29.4%) for BCS and mastectomy, respectively. Figure 2A shows the Nelson-Aalen cumulative-hazard rates for LRI by surgical type. There was no significant difference in the estimated 18-month LRR between mastectomy and BCS (hazard ratio, HR: 1.43; 95% confidence interval, CI: 0.89–2.32; P = 0.143). However, patients undergoing BCS had a significantly higher LRR at 5 years (2.63% vs 5.33%; HR, 3.39; 95% CI, 2.03–5.66; P < 0.001) and at 10 years (4.93% vs 11.68%; HR, 5.27; 95% CI, 2.43–11.43; P < 0.001). The change in HR over time is illustrated in Fig. 2B, with the HR crossing one at 12 months. Similar results were found when excluding patients with a maximum overall (invasive + in situ) tumor diameter >30 mm (thus excluding patients with larger tumors who were likely to be treated with a mastectomy) (data not shown). MVA showed that patients undergoing BCS also had a significantly higher chance of a local recurrence at both 5 and 10 years, with only adjuvant radiotherapy significantly affecting the MVA (Table 2). When looking at LRI for mastectomy by CWR-XRT (Supplementary Figure 1A), patients without CWR-XRT had a significantly higher LRR compared with those with CWR-XRT (HR, 0.46; 95% CI, 0.24–0.86; P = 0.015). However, when assessing LRI for BCS patients by radiotherapy boost, no significant differences were found between patients with/without a boost (HR, 0.90; 95% CI, 0.58–1.38; P = 0.614) (Supplementary Figure 1B).

RT had a significantly higher LRR compared with those with CWR-XRT (HR, 0.46; 95% CI, 0.24–0.86; P = 0.015). However, when assessing LRI for BCS patients by radiotherapy boost, no significant differences were found between patients with/without a boost (HR, 0.90; 95% CI, 0.58–1.38; P = 0.614) (Supplementary Figure 1B). There was also no difference for BCS for those with surgical margins of 0 mm versus >0 mm in terms of LRI (HR, 0.86; 95% CI, 0.41–1.78; P = 0.680) (Supplementary Figure 1C). FIGURE 2 Local-recurrence interval for all patients by surgical type. A, Nelson-Aalen cumulative hazard plot. B, Flexible parametric survival model time-varying hazard over time. TABLE 2 Local Recurrence Interval Flexible Parametric Survival Model Multivariable Analysis Results for all Patients (Excluding Those With Neoadjuvant Chemotherapy)

FIGURE 2 Local-recurrence interval for all patients by surgical type. A, Nelson-Aalen cumulative hazard plot. B, Flexible parametric survival model time-varying hazard over time. TABLE 2 Local Recurrence Interval Flexible Parametric Survival Model Multivariable Analysis Results for all Patients (Excluding Those With Neoadjuvant Chemotherapy) Covariate HR* 95% CI P Surgical type at 5 years Mastectomy 1 (Ref. cat.) — — BCS (unadjusted) 5.33 4.15 to 6.85 <0.001 BCS (adjusted) 5.00 3.57 to 23.69 <0.001 Surgical type at 10 years Mastectomy 1 (Ref. cat.) — — BCS (unadjusted) 11.68 9.05 to 15.07 <0.001 BCS (adjusted) 6.06 1.29 to 28.40 0.022 Age at diagnosis, y, (continuous) 1.02 0.95 to 1.09 0.662 Maximum overall (invasive + in situ) tumor size, (mm) (continuous) 1.42 0.78 to 2.58 0.253 Focality Localized 1 (Ref. cat.) — — Multifocal 1.15 0.58 to 2.30 0.688 N stage N0 1 (Ref. cat.) — — N1 1.18 0.71 to 1.96 0.527 Histological grade 1 1 (Ref. cat.) — — 2 1.63 0.38 to 7.02 0.514 3 1.42 0.33 to 6.16 0.636 ER status Negative 1 (Ref. cat.) — — Positive 0.64 0.28 to 1.48 0.297 HER2 status Negative 1 (Ref. cat.) — — Positive 1.33 0.77 to 2.30 0.306 Adjuvant radiotherapy No/unknown 1 (Ref. cat.) — — Yes 0.32 0.16 to 0.64 0.001 Adjuvant hormone therapy No/unknown 1 (Ref. cat.) — — Yes 0.64 0.28 to 1.47 0.295 Surgical margins, mm 0 1 (Ref. cat.) — — ≥0 to <1 0.74 0.26 to 2.14 0.579 ≥1 to ≤5 0.93 0.40 to 2.18 0.871 >5 0.89 0.36 to 2.22 0.803 CI indicates confidence interval; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; HR, hazard ratio.

herapy No/unknown 1 (Ref. cat.) — — Yes 0.64 0.28 to 1.47 0.295 Surgical margins, mm 0 1 (Ref. cat.) — — ≥0 to <1 0.74 0.26 to 2.14 0.579 ≥1 to ≤5 0.93 0.40 to 2.18 0.871 >5 0.89 0.36 to 2.22 0.803 CI indicates confidence interval; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; HR, hazard ratio. *Unless otherwise stated, HR presented for the multivariable (adjusted) model.

herapy No/unknown 1 (Ref. cat.) — — Yes 0.64 0.28 to 1.47 0.295 Surgical margins, mm 0 1 (Ref. cat.) — — ≥0 to <1 0.74 0.26 to 2.14 0.579 ≥1 to ≤5 0.93 0.40 to 2.18 0.871 >5 0.89 0.36 to 2.22 0.803 CI indicates confidence interval; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; HR, hazard ratio. *Unless otherwise stated, HR presented for the multivariable (adjusted) model. DDFI by surgical type showed that mastectomy patients had a significantly worse DDFI than BCS (HR, 0.51; 95% CI, 0.44–0.60; P < 0.001) (Fig. 3). However, in the MVA the difference was no longer significant (HR, 0.82; 95% CI 0.64–1.05; P = 0.115) (Supplementary Table 1). Factors affecting the MVA were maximum invasive tumor size, N stage, grade, and ER status. DDFI by patients experiencing versus not experiencing a local-recurrence event identified that DDFI was similar at 5 years but the hazards separated at 10 years (HR, 0.80; 95% CI 0.54–1.18; P = 0.263 and HR, 0.29; 95% CI 0.14–0.62; P = 0.001, respectively) (Supplementary Figure 2A). When assessing DDFI by BCS patients with surgical margins of 0 mm versus >0 mm, those with margins >0 mm had a significantly better DDFI compared with those with 0 mm margins (Supplementary Figure 2B). Similar results were also found in OS. UVA of OS by surgical type demonstrated that mastectomy patients had a significantly worse OS compared with BCS (HR, 0.53; 95% CI 0.45–0.62; P < 0.001) (Fig. 4), and in the MVA the difference was no longer significant (HR, 0.79; 95% CI 0.61–1.03; P = 0.081) (Supplementary Table 2). Excluding ER status, the same factors also affected the MVA for OS. Moreover, when looking at OS by local-recurrence event and by surgical margins, the findings matched those of the DDFI analyses (Supplementary Figure 3A and 3B).

was no longer significant (HR, 0.79; 95% CI 0.61–1.03; P = 0.081) (Supplementary Table 2). Excluding ER status, the same factors also affected the MVA for OS. Moreover, when looking at OS by local-recurrence event and by surgical margins, the findings matched those of the DDFI analyses (Supplementary Figure 3A and 3B). FIGURE 3 Distant disease free interval Kaplan-Meier plot for all patients by surgical type. FIGURE 4 Overall survival Kaplan-Meier plot for all patients by surgical type. DISCUSSION Previous findings from the POSH study reported on the effects of ethnicity and obesity, both of which have affected outcome including DDFI in this young age group,27,28 and family history, which has not.17 This study has investigated the effect of surgery on LRI, DDFI, and OS, and the effect of local recurrence on DDFI and OS in young women with breast cancer. A large number of studies use inconsistent definitions of local-recurrence, often not specifying which events have been included in the local-recurrence definition.29 This study has therefore clearly described the definition of local recurrence in the methods with criteria outlining which events were included/excluded depending on the time of competing events. This study also incorporated the use of FPSMs to assess the time-varying effect of surgical type on LRI.

rence definition.29 This study has therefore clearly described the definition of local recurrence in the methods with criteria outlining which events were included/excluded depending on the time of competing events. This study also incorporated the use of FPSMs to assess the time-varying effect of surgical type on LRI. Previous findings from a meta-analysis of RCTs conducted by the EBCTCG18 presented first recurrence rates (local and distant) which appeared much higher after BCS in younger women; 36.1% for women <40 years undergoing BCS with radiotherapy. However, the number analyzed was relatively small (n = 363) and there was no breakdown in the meta-analysis of local versus distant recurrences. In this study, the majority of events were found to be distant and not local (139 LRI vs 752 DDFI events) which demonstrates the predominant risk in these young patients is of distant and not local recurrence.

elatively small (n = 363) and there was no breakdown in the meta-analysis of local versus distant recurrences. In this study, the majority of events were found to be distant and not local (139 LRI vs 752 DDFI events) which demonstrates the predominant risk in these young patients is of distant and not local recurrence. Although UVA demonstrated worse DDFI and OS for mastectomy compared with BCS, this is almost certainly because of imbalances in prognostic features between the groups. Patients undergoing mastectomy had significantly larger tumors than BCS, with a higher proportion of tumors EIC+, and ER+, PR+, and/or HER2+. Unsurprisingly, tumors of maximum invasive tumor size >30 mm, N1 stage, and grade 3 were significant factors in both DDFI and OS MVA, the differences between surgical type for DDFI and OS were no longer significant after correction for these factors. Interestingly, maximum overall (invasive + in situ) tumor size >30 mm was not a significant factor in either OS or DDFI MVA models, indicating that whereas overall tumor size influences surgical decision making, invasive tumor size is the relevant size parameter predicting DDFI and OS. A sensitivity analysis excluding patients with a maximum overall tumor size >30 mm showed similar results in a UVA of LRI comparing surgical type.

r DDFI MVA models, indicating that whereas overall tumor size influences surgical decision making, invasive tumor size is the relevant size parameter predicting DDFI and OS. A sensitivity analysis excluding patients with a maximum overall tumor size >30 mm showed similar results in a UVA of LRI comparing surgical type. The results of this study support existing literature in regards to both OS and DDFI by surgical type, with no evidence of surgical type affecting survival or distant-disease in this age group. This is consistent with RCTs comparing mastectomy versus BCS in the breast cancer population as a whole4–6,11 indicating that surgical choices for younger women can use accepted criteria without impacting outcome. The results of this study also demonstrated similar LRR in the first 18 months for mastectomy and BCS, but a larger disparity is seen at 5 and 10 years, with significantly higher LRR for BCS. The clinical implication is that, at least initially, for local recurrence there is no disadvantage in treating young women surgically with BCS in general and no evidence that BCS leads to a disadvantage in DDFI or OS. It is not yet possible to comment beyond 10 years at this stage for this cohort. However, other studies suggest LRR continue to rise beyond 10 years after BCS.19,30

recurrence there is no disadvantage in treating young women surgically with BCS in general and no evidence that BCS leads to a disadvantage in DDFI or OS. It is not yet possible to comment beyond 10 years at this stage for this cohort. However, other studies suggest LRR continue to rise beyond 10 years after BCS.19,30 An interesting finding from this study is the effect of margin on outcomes. Although no effect was seen for LRI, differences were observed for both OS and DDFI; positive margins were associated with significantly worse OS and DDFI. This could be because of reduced power of LRI outcome because of a fewer local-recurrence events and missing data for margin status, or possibly because of patients with a positive margin being more likely to present with a distant relapse, or combination of distant disease and local recurrence, which would, as defined here, be considered a distant event, as these would not be surgically salvageable, isolated local recurrences. In this analysis, a positive margin was defined according to ASCO guidance as tumor at the margin.24 Although 24.5% of margin status information was missing, sensitivity analyses of MVA models using multiple imputation were carried out and showed very similar results to the complete-case analyses. The finding that surgical margins are a factor in the development of distant disease would support the concept of the importance of surgical quality with attention to margins, with re-excision where appropriate. Taken together with the lack of evidence here that oncological surgical type influences distant-relapse it could be argued that completeness of excision is more important than the extent of surgery.

support the concept of the importance of surgical quality with attention to margins, with re-excision where appropriate. Taken together with the lack of evidence here that oncological surgical type influences distant-relapse it could be argued that completeness of excision is more important than the extent of surgery. In regards to radiotherapy, patients treated with BCS who were not documented to have adjuvant radiotherapy unsurprisingly had higher LRR, implying that the data were correct (rather than data missing because of patients having radiotherapy elsewhere), and highlighting the importance of radiotherapy as part of breast conservation. Although no effect of radiotherapy boost was shown for patients undergoing BCS, it must be noted that this is not an RCT. Interestingly, provision for radiotherapy boost was not shown to be statistically correlated with margin status; however, the clinical implication we would draw is that, at least in this study, provision of a radiotherapy boost appears to be less important than attention to detail to surgical margins in terms of its effect on LRI and DDFI.

ion for radiotherapy boost was not shown to be statistically correlated with margin status; however, the clinical implication we would draw is that, at least in this study, provision of a radiotherapy boost appears to be less important than attention to detail to surgical margins in terms of its effect on LRI and DDFI. More than 60% of patients undergoing mastectomy received CWR-XRT and a clear association of benefit of CWR-XRT on LRI has been demonstrated here, despite potential confounding. Given results of the most recent Oxford overview31 it is likely that thresholds for CWR-XRT after mastectomy are likely to fall further. Given that the majority of young patients are likely to receive radiotherapy, even if surgically treated with mastectomy, there are likely to be implications for reconstructive decision making. These findings also support the message that avoiding local recurrence is important as increased local recurrence is associated with poorer DDFI and OS.30,32,33 Our data suggest that valid strategies to reduce local recurrence might include avoidance of a positive margin after BCS and provision of CWR-XRT after mastectomy where indicated, but do not support mastectomy over BCS where both options are available.

sed local recurrence is associated with poorer DDFI and OS.30,32,33 Our data suggest that valid strategies to reduce local recurrence might include avoidance of a positive margin after BCS and provision of CWR-XRT after mastectomy where indicated, but do not support mastectomy over BCS where both options are available. In this analysis, the frequency of local recurrence is much lower than that of distant relapse indicating that the main hazard experienced by these patients, at least within the first 10 years, is of distant rather than local recurrence. We have not demonstrated an impact of tumor stage or biological type on local recurrence in this analysis, possibly because of reduced power because of a lower LRR. In addition to young age, factors recognized to influence local recurrence after BCS and mastectomy include axillary nodal status, margin status, and lack of systemic therapy.34 When considering molecular subtype a greater proportion of young women appear to have luminal B tumors2; however, young age remains predictive of LRR independent of molecular subtype,35 although there is a suggestion that molecular subtype may affect local recurrence.36 Younger patients with breast cancer are also more likely to carry a germline BRCA-mutation and it is currently unknown whether this influences LRI or DDFI, although clearly it does increase second new primary breast tumors; and contralateral new primary events were not included in this analysis of local recurrence. Once final genotyping in this cohort has been completed, further analyses will also be performed by BRCA status to see if this has an effect.

nces LRI or DDFI, although clearly it does increase second new primary breast tumors; and contralateral new primary events were not included in this analysis of local recurrence. Once final genotyping in this cohort has been completed, further analyses will also be performed by BRCA status to see if this has an effect. Regardless of this, the current Association of Breast Surgery guidelines state that the target local-recurrence rate after surgery should be <3% and not >5% at 5 years.37 This study has demonstrated that LRI in younger patients treated by mastectomy would fulfill this criterion (HR, 2.63; 95% CI 1.85–3.74), and that the lower 95% LRI limit for younger women undergoing BCS is within this range (HR, 5.33; 95% CI 4.15–6.85).37 Furthermore, our findings are consistent with recommendations for breast surgery within recent consensus guidelines for the management of young women with breast cancer.38 A limitation of this study is that as this was not an RCT, any differences/lack of differences in LRI, OS, and/or DDFI were because of the surgical type alone could be the result of confounding. However, we have accounted as far as possible for biases and this is a large prospective cohort representative of cancer treatment in this age group in the United Kingdom.23 It should be noted this analysis was performed according to a prespecified plan and LRI was clearly defined to address the inconsistency of reporting in a number of previous studies.29

ossible for biases and this is a large prospective cohort representative of cancer treatment in this age group in the United Kingdom.23 It should be noted this analysis was performed according to a prespecified plan and LRI was clearly defined to address the inconsistency of reporting in a number of previous studies.29 In conclusion, there are no survival advantages for surgical type after adjusting for known prognostic factors. There is no difference in LRI between BCS and mastectomy in young women with breast cancer in the short-term but, beyond 18 months, LRR are higher after BCS. Local recurrence is associated with increased risk of distant relapse, and in patients undergoing BCS, a positive surgical margin increases the likelihood of a distant relapse. For those undergoing mastectomy, CWR-XRT reduces the likelihood of distant relapse. Surgical extent therefore appears less important for DDFI than completeness of excision or, where appropriate, CWR-XRT. Future work will assess the impact of germline genotype on LRI, distant relapse, and OS.

ihood of a distant relapse. For those undergoing mastectomy, CWR-XRT reduces the likelihood of distant relapse. Surgical extent therefore appears less important for DDFI than completeness of excision or, where appropriate, CWR-XRT. Future work will assess the impact of germline genotype on LRI, distant relapse, and OS. Supplementary Material Supplemental Digital Content Acknowledgments The authors would like to thank all the patients who participated in this study and the POSH collaborators including the POSH Steering group: Professor Diana Eccles, Dr. Peter Simmonds, Professor Douglas G. Altman, Professor Paul Pharoah, Professor Louise Jones, Professor Ros Eeles, Professor D. Gareth Evans, Professor Andrew Hanby, Professor Alistair M. Thompson, Professor Shirley Hodgson, Mr Hisham Hamed, Dr Fiona Gilbert, and Professor Sunil Lakhani. Participating principal investigators are listed on the study website: http://www.southampton.ac.uk/medicine/research/posh.page. We thank the NIHR NCRN for supporting patient recruitment, and all patients participating in the POSH study. Tom Maishman and Ramsey I Cutress contributed equally to this work and so are joint first authors. Disclosure: E.R.C. has received honoraria from Roche. Funding for data collection and analysis for the POSH study was provided by Cancer Research UK (grants A7572, A11699, and C1275/A15956) and the study was sponsored by University Hospital Southampton NHS Foundation Trust. The authors declare no conflict of interest.

Esophageal adenocarcinoma (EADC) incidence is increasing1 and currently represents 5% of the digestive tract cancers in the UK.2 Overall disease survival is poor,3 but correlates with stage of cancer at presentation, demonstrating significant survival advantages with detection of early-stage disease.4,5 Barrett esophagus (BE), in which normal squamous mucosa is replaced with a metaplastic columnar phenotype, results from prolonged exposure to stomach acids and bile salts, which reflux into the esophagus causing chronic inflammation and tissue damage.6 The incidence of BE is increasing, largely thought to be a consequence of obesity-induced reflux disease.7–10 BE is associated with an increased risk of EADC,11 but for the majority of patients, BE will never progress beyond simple benign metaplasia.12,13 However, in a small number of patients, dysplasia will develop, with some progressing to EADC.14 The incidence of EADC in the BE population is up to 150 times greater than unaffected individuals.12 Although the pathological changes seen in Barrett adenocarcinoma are understood as part of a well-established metaplasia-dysplasia-carcinoma sequence,15 the molecular drivers are less clear.16–18 The current dilemma is that for patients with nondysplastic BE, there are no accurate methods for identifying the small number of patients at high risk of progression to cancer.

nocarcinoma are understood as part of a well-established metaplasia-dysplasia-carcinoma sequence,15 the molecular drivers are less clear.16–18 The current dilemma is that for patients with nondysplastic BE, there are no accurate methods for identifying the small number of patients at high risk of progression to cancer. Surveillance practice of the Barrett patients varies widely, between some who endoscope patients each year in contrast to others who will never repeat the investigation.19 The ongoing UK Medical Research Council funded BOSS study aims to understanding the optimum surveillance strategy, randomizing between prospective monitoring BE patients with frequent endoscopic assessment or a “watch and wait” policy.20

atients each year in contrast to others who will never repeat the investigation.19 The ongoing UK Medical Research Council funded BOSS study aims to understanding the optimum surveillance strategy, randomizing between prospective monitoring BE patients with frequent endoscopic assessment or a “watch and wait” policy.20 Clearly, there is need for a method of risk stratification in these patients to facilitate a streamlined surveillance program by identifying high-risk nondysplastic BE patients. Attempts at biomarker development for stratification of high-risk BE have focused on mutational change, specifically around the role of TP53 mutation in predicting “high-risk” disease,21 given its role as a driver in esophageal cancer. However, Ross-Innes et al17 have convincingly demonstrated the presence of pathogenic TP53 mutations in apparently normal squamous esophageal mucosa, thus making its role in progression to invasive adenocarcinoma unclear. However, the role of epigenetic change in the pathogenesis of BE and esophageal cancer is less well-understood, but may well happen much earlier in the cancer development pathway, and, as a direct result, provide a more appropriate target for both predicting its development and potentially arresting tumorigenesis, should a suitable epigenetic modulator be identified.

hogenesis of BE and esophageal cancer is less well-understood, but may well happen much earlier in the cancer development pathway, and, as a direct result, provide a more appropriate target for both predicting its development and potentially arresting tumorigenesis, should a suitable epigenetic modulator be identified. Multiple methylation markers have been identified which can discriminate between high-risk and low-risk BE including APC/p16,22MGMT,23PKP-1,24TIMP3/TERT,25RUNX3/HPP1,26,27 and AKAP12.28 Agarwal et al29 performed a MeCIP array-based approach to compare the methylomes of progressor (n = 5) versus nonprogressing patients (n = 4). In patients who progressed to invasive adenocarcinoma, their original biopsies began either with no dysplasia, indefinite dysplasia, or low-grade dysplasia, making comparison difficult. However, subsequent analysis of the top 25 differential methylation patterns found 3 gene regions with hypermethylation amongst the progression group (Pro_MMD2, Pro_ZNF358, and Intra_F10), with a trend towards global hypomethylation, in keeping with other epithelial premalignant conditions.30 Kaz et al31 also found significant differences in methylation in patients with BE due to factors such as obesity, smoking, and sex, which may be responsible for some of the observed risk. While the studies reviewed do show variation in methylation between progressive versus nonprogressive BE, the methodology has been heterogeneous, and few have conducted the study with a group of the same patients tracked over time.

Multiple methylation markers have been identified which can discriminate between high-risk and low-risk BE including APC/p16,22MGMT,23PKP-1,24TIMP3/TERT,25RUNX3/HPP1,26,27 and AKAP12.28 Agarwal et al29 performed a MeCIP array-based approach to compare the methylomes of progressor (n = 5) versus nonprogressing patients (n = 4). In patients who progressed to invasive adenocarcinoma, their original biopsies began either with no dysplasia, indefinite dysplasia, or low-grade dysplasia, making comparison difficult. However, subsequent analysis of the top 25 differential methylation patterns found 3 gene regions with hypermethylation amongst the progression group (Pro_MMD2, Pro_ZNF358, and Intra_F10), with a trend towards global hypomethylation, in keeping with other epithelial premalignant conditions.30 Kaz et al31 also found significant differences in methylation in patients with BE due to factors such as obesity, smoking, and sex, which may be responsible for some of the observed risk. While the studies reviewed do show variation in methylation between progressive versus nonprogressive BE, the methodology has been heterogeneous, and few have conducted the study with a group of the same patients tracked over time. The study aims to determine whether there are differences in methylation in patients between high-risk nondysplastic BE, which will progress to cancer, versus low-risk BE.

While the studies reviewed do show variation in methylation between progressive versus nonprogressive BE, the methodology has been heterogeneous, and few have conducted the study with a group of the same patients tracked over time. The study aims to determine whether there are differences in methylation in patients between high-risk nondysplastic BE, which will progress to cancer, versus low-risk BE. METHODS Patients and Samples Two sample cohorts were identified containing patients who either progressed to EADC from nondysplastic BE or remained with nondysplastic BE (NDBE) identified from a prospectively maintained database of patients with BE at a large district general hospital. Inclusion criteria for the study were progressing patients with nondysplastic BE who, when observed over the study period, developed EADC. Samples were only included where there was a NDBE biopsy and then histology evidence that the patient developed adenocarcinoma. Nonprogressing patients were identified from a biopsy of NDBE, which, when followed over time, never progressed beyond NDBE. To be included in this group, the patient must have been in a surveillance program for a minimum of 15 years and have serial biopsies over that period. Patients within the surveillance program had endoscopy and biopsy every 2 years.32 Biopsies were taken at the time of the initial surveillance endoscopy and at all subsequent endoscopies including immediately before treatment as part of their staging. They were also required to have still been alive and to have had a NDBE biopsy within 2 years of this study.

had endoscopy and biopsy every 2 years.32 Biopsies were taken at the time of the initial surveillance endoscopy and at all subsequent endoscopies including immediately before treatment as part of their staging. They were also required to have still been alive and to have had a NDBE biopsy within 2 years of this study. Patients were excluded from the study if BE material was only available as part of tumor-associated BE or if dysplasia was identified in any BE biopsies. All tissue used was formalin-fixed paraffin-embedded (FFPE) samples obtained from pathology libraries and prepared by the University of Birmingham Human Biomaterials Resource Centre (ethical approval 09/H1010/75). H&E stained slides were reviewed by a consultant pathologist to ensure that the samples were BE and had no dysplasia throughout their extent. Cut 5-μM paraffin sections were mounted onto frosted slides, and macrodissection for BE was carried out. DNA extraction was then performed using Qiagen DNeasy Blood and Tissue kit following the manufacturer's protocol. Each sample of extracted DNA was then quantified and qualified by Nanodrop spectrophotometry and Qubit fluorimetry. Bisulfite conversion was performed using a Zymo DNA methylation bisulfite conversion kit following the modified Illumina Infinium protocol on 500 ng of extracted DNA.

ollowing the manufacturer's protocol. Each sample of extracted DNA was then quantified and qualified by Nanodrop spectrophotometry and Qubit fluorimetry. Bisulfite conversion was performed using a Zymo DNA methylation bisulfite conversion kit following the modified Illumina Infinium protocol on 500 ng of extracted DNA. Methylation Arrays The Illumina HumanMethylation 450 array, in which the methylation status of more than 485,000 individual CpG sites are examined,33 was used to compare sample groups (progressing NDBE vs nonprogressing NDBE). Once bisulfite converted, 1 ng of DNA was quality-controlled (Illumina FFPE QC kit) with only samples with dCt <5 being taken forward to array analysis. The resulting samples underwent repair suitable for array hybridization using the Illumina FPPE restore kit,34 followed by hybridization to Illumina HumanMethylation450 arrays using manufacturer's protocols and scanned on an Illumina iScan. Normalized intensity files (iDAT) were exported using GenomeStudio for downstream analysis. Immunohistochemistry Immunohistochemistry (IHC) was carried out on a Leica Bond RX system using a mouse polyclonal anti-OR3A4 antibody (Abcam ab67107) at a dilution of 1:100 with a primary incubation time of 15 minutes.

Methylation Arrays The Illumina HumanMethylation 450 array, in which the methylation status of more than 485,000 individual CpG sites are examined,33 was used to compare sample groups (progressing NDBE vs nonprogressing NDBE). Once bisulfite converted, 1 ng of DNA was quality-controlled (Illumina FFPE QC kit) with only samples with dCt <5 being taken forward to array analysis. The resulting samples underwent repair suitable for array hybridization using the Illumina FPPE restore kit,34 followed by hybridization to Illumina HumanMethylation450 arrays using manufacturer's protocols and scanned on an Illumina iScan. Normalized intensity files (iDAT) were exported using GenomeStudio for downstream analysis. Immunohistochemistry Immunohistochemistry (IHC) was carried out on a Leica Bond RX system using a mouse polyclonal anti-OR3A4 antibody (Abcam ab67107) at a dilution of 1:100 with a primary incubation time of 15 minutes. Immunohistochemistry was scored on epithelial and stromal components and a composite score consisting of the sum of expression within membranous, nuclear and cytoplasmic compartments on a score of 1 to 4 was made, giving a combined maximum possible score of 12 for each compartment. Scoring was carried out by 2 independent observers blinded to progressor/nonprogressor status.

components and a composite score consisting of the sum of expression within membranous, nuclear and cytoplasmic compartments on a score of 1 to 4 was made, giving a combined maximum possible score of 12 for each compartment. Scoring was carried out by 2 independent observers blinded to progressor/nonprogressor status. Bioinformatics Analysis of Array Data Bioinformatics analysis of the methylation microarrays was carried using the ChAMP package35 via Bioconductor/R. In brief, red/green intensity values were captured from Illumina iDAT files, background-corrected and SWAN-normalized to produce M values (further details given in supplementary methods). M values were analyzed using a logistic regression model using Empirical Bayesian shrinkage of moderated t-statistics to correct for small sample size. Small sample size was controlled for by setting stringent false discovery rate Q values of <0.05. Identification of variable methylated sites allowed the CpG site markers to be highlighted. DNA copy number analysis was carried out using the DNAcopy module of ChAMP.

derated t-statistics to correct for small sample size. Small sample size was controlled for by setting stringent false discovery rate Q values of <0.05. Identification of variable methylated sites allowed the CpG site markers to be highlighted. DNA copy number analysis was carried out using the DNAcopy module of ChAMP. Pyrosequencing Validation of Hits Methylation-insensitive primers were designed and sourced, using Qiagen PyroMark Primer Design software v2.0. Primers were designed to flank CpG sites of interest. Illumina CG methylation probe locations were retrieved from the UCSC genome browser,36 and Fast-All sequence retrieved for −200 bp to +200 bp of the target CG dinucleotide. Primer design settings were optimized to design amplicons suitable for FFPE pyrosequencing, with the optimum amplicon size set to between 80 and 150 bp. All other settings were as per the standard Qiagen design parameters. Primers were ordered from Sigma-Aldrich, with the biotinylated pyrosequencing primer being purified by high-performance liquid chromatography and the remainder by desalting. Pyrosequencing PCR was performed using Qiagen PyroMark PCR Gold kit, consisting of 2 μL of bisulfite-converted DNA, 25 μL of PCR master mix, 5 μL of CoralLoad dye, 3 μL MgSO4, 10 μL of Q reagent, and 2.5 μL each of forward (20 mM) and reverse (20 mM) primer. Reaction conditions were determined experimentally by use of a gradient PCR for each primer pair. A typical reaction consisted of activation at 95C for 15 minutes, followed by 45 cycles of denaturation at 94°C for 30 seconds, annealing at 56°C for 30 seconds, and extension at 72°C for 30 seconds, followed by a final extension step at 72°C for 10 minutes. In addition to experimental DNA, each PCR was performed with 100% methylated DNA, 100% unmethylated DNA, and ddH2O as controls. Methylated and unmethylated DNA was generated in house by means of M.SSl conversion (methylated DNA) and whole genome amplification using the Qiagen Repli-G kit (unmethylated DNA).

10 minutes. In addition to experimental DNA, each PCR was performed with 100% methylated DNA, 100% unmethylated DNA, and ddH2O as controls. Methylated and unmethylated DNA was generated in house by means of M.SSl conversion (methylated DNA) and whole genome amplification using the Qiagen Repli-G kit (unmethylated DNA). Primer sequences for validation pyrosequencing were as follows: for FGFR2 cg17337672 these were forward = AGGGGAAGGGAATTTAGGTT, reverse = [Btn]TCAATCTTCCCCCAAACAACCACT, and sequencing = GTTTAGAAGTTTTTTTTGGATTAGT; for ORA3A4 cg07863524 these were forward = GTGGTAGAAGTAGGATGAGGTGTTGATAAT, reverse = [Btn]CTTCAACTTCCTTCCCCTTACATTT, and sequencing = GGGTAGGGATGGAAGA; for OR3A4 cg09890332 these were forward = TTAAAGTGTTAGGATTATAGGTGTGAGTTA, reverse = [Btn]TTTCCCAACCCTAATCACTACTAATAAAAT, and sequencing = GGATTATAGGTGTGAGTTAT.

GATTAGT; for ORA3A4 cg07863524 these were forward = GTGGTAGAAGTAGGATGAGGTGTTGATAAT, reverse = [Btn]CTTCAACTTCCTTCCCCTTACATTT, and sequencing = GGGTAGGGATGGAAGA; for OR3A4 cg09890332 these were forward = TTAAAGTGTTAGGATTATAGGTGTGAGTTA, reverse = [Btn]TTTCCCAACCCTAATCACTACTAATAAAAT, and sequencing = GGATTATAGGTGTGAGTTAT. RESULTS Patient Selection In all, 67 patients were recruited—37 from Sandwell and West Birmingham NHS Trust (SWBNT) and 30 from University Hospital Birmingham NHS Trust (UHBFT). Of these, 20/67 progressed from nondysplastic BE and 47 did not. The age range was between 42 and 60 years, with a median age of 56 years, of which 60/67 (89.6%) were male. The median time to diagnosis of EADC in “progressor” patients was 114 months, with a range of 14–162 months. Of the patients recruited, in the SWBNT group, 6 progressors and 6 nonprogressors, and in the UHBFT, 6 progressors and 6 nonprogressors were taken forward to methylation array analysis, giving a total of 12 progressors and 12 nonprogressors. This samples size was chosen because of our previous experience with biomarker discovery in methylation arrays as a suitable size for biomarker discovery. A validation cohort of 32 patients (progressors 18, nonprogressors 14) were obtained from University Hospital Southampton. All patients included in the study had symptoms of reflux disease as a presenting symptom. For all patients in the progressor cohort (n = 30), the observed pathological disease stages at the time of resection were high-grade dysplasia (4/30, 13%), T1N0 (10/30, 33%), T1N1 (1/30, 3%), T2N0 (2/30, 7%), T2N1 (3/30,10%), T3N0 (6/30, 20%), and T3N1 (4/30, 13%). All patients with high-grade dysplasia underwent endoscopic mucosal resection and the remainder underwent esophagectomy. All patients recruited had validation pyrosequencing performed.

de dysplasia (4/30, 13%), T1N0 (10/30, 33%), T1N1 (1/30, 3%), T2N0 (2/30, 7%), T2N1 (3/30,10%), T3N0 (6/30, 20%), and T3N1 (4/30, 13%). All patients with high-grade dysplasia underwent endoscopic mucosal resection and the remainder underwent esophagectomy. All patients recruited had validation pyrosequencing performed. Methylation Microarray Analysis Twenty-four samples in total were hybridized successfully to Illumina HumanMethylation450 microarrays. All arrays passed manufacturers QC as specified by metrics in Illumina GenomeStudio. Differential methylation analysis at the probe level (Table 1) revealed significant differences in methylation between progressor and nonprogressors in nondysplastic BE (Fig. 1). In all, 44 significantly [defined as Bayes factor (BF) >5, chosen as it is equivalent to a genome, wide P value significance of 1 × 10−6] differentially methylated targets were identified, the bulk being hypomethylated, with a trend towards global hypomethylation in progressor samples as demonstrated by left-shift of the Volcano plot (Fig. 1). TABLE 1 Top 20 Array-identified CpG Sites With Methylation Variation Between Nondysplastic Samples Which Progress to EADC Versus Those Which Remain Static

Methylation Microarray Analysis Twenty-four samples in total were hybridized successfully to Illumina HumanMethylation450 microarrays. All arrays passed manufacturers QC as specified by metrics in Illumina GenomeStudio. Differential methylation analysis at the probe level (Table 1) revealed significant differences in methylation between progressor and nonprogressors in nondysplastic BE (Fig. 1). In all, 44 significantly [defined as Bayes factor (BF) >5, chosen as it is equivalent to a genome, wide P value significance of 1 × 10−6] differentially methylated targets were identified, the bulk being hypomethylated, with a trend towards global hypomethylation in progressor samples as demonstrated by left-shift of the Volcano plot (Fig. 1). TABLE 1 Top 20 Array-identified CpG Sites With Methylation Variation Between Nondysplastic Samples Which Progress to EADC Versus Those Which Remain Static Probe ID t P Adj. P Gene Name cg09890332 −12.26 2.02E-08 0.0031 OR3A4 cg24007926 −11.05 6.84E-08 0.0032 NA cg17337672 −10.95 7.54E-08 0.0032 FGFR2 cg02226469 −10.88 8.17E-08 0.0032 NA cg17433294 −10.51 1.22E-07 0.0038 NMUR2 cg18479711 −10.31 1.52E-07 0.0039 HDAC4 cg09011162 −10.15 1.82E-07 0.0040 LMF1 cg19733463 −9.83 2.61E-07 0.0045 NMUR1 cg16150571 −9.64 3.26E-07 0.0045 SNORD116-22 cg24424217 −9.62 3.35E-07 0.0045 ZNF511 cg13164993 −9.57 3.52E-07 0.0045 RBP3 cg14019464 −9.53 3.70E-07 0.0045 TRIB3 cg24581378 −9.53 3.71E-07 0.0045 ZAP70 cg05230642 −9.36 4.54E-07 0.0051 SNORD115-14 cg12297814 −9.26 5.08E-07 0.0052 IGFN1 cg11231240 −9.23 5.33E-07 0.0052 NA cg11864327 −8.96 7.40E-07 0.0063 ZFP2 cg16771467 −8.92 7.78E-07 0.0063 ATP8B1 cg17304276 −8.88 8.14E-07 0.0063 CUX2 cg11443888 −8.86 8.36E-07 0.0063 TMEM151B Probe ID: the Illumina cg probe ID from the Illumina manifest; T: the t value (the size of the difference relative to the variation in the sample).

E-07 0.0052 NA cg11864327 −8.96 7.40E-07 0.0063 ZFP2 cg16771467 −8.92 7.78E-07 0.0063 ATP8B1 cg17304276 −8.88 8.14E-07 0.0063 CUX2 cg11443888 −8.86 8.36E-07 0.0063 TMEM151B Probe ID: the Illumina cg probe ID from the Illumina manifest; T: the t value (the size of the difference relative to the variation in the sample). P = the raw P value, not corrected for multiple testing; adjusted P = the P value corrected for multiple testing. FIGURE 1 Volcano plot of probe-level methylation in progressors versus nonprogressors. Blue points = Bayes factor <5; red points = Bayes factor >5. The plot shows a leftward shift of probes towards the left, suggesting global hypomethylation. Differential Methylation at the Probe Level The top ranked differentially methylated probe was cg09890332 (chr17:3212495–3212495, hg19 coordinates), which tags a CpG dinucleotide −1044 bp upstream of the transcription start site of the long noncoding RNA, OR3A4 (NRR_024128.1). The second highest ranked differentially methylated probe was cg24007926 (chr2:206842761–206842761, hg19 coordinates). This CpG dinucleotide is within a large, intragenic region, with the nearest gene being INO80D (INO80 complex subunit D, NM_017759.4), 15,684 bp downstream of this CpG. The third highest ranked differentially methylated probe was cg17337672 (chr10:123354172–123354172), which tags a CpG dinucleotide within intron 2 of FGFR2 (fibroblast growth factor receptor 2, NM_000141.4).

on, with the nearest gene being INO80D (INO80 complex subunit D, NM_017759.4), 15,684 bp downstream of this CpG. The third highest ranked differentially methylated probe was cg17337672 (chr10:123354172–123354172), which tags a CpG dinucleotide within intron 2 of FGFR2 (fibroblast growth factor receptor 2, NM_000141.4). Differentially methylated regions (DMRs) were called between progressors and nonprogressors via the dmrLasso function of the CHAMP software package (Table 2). Significant DMRs were found from chr2:503065–503193 (which tags an intragenic region, DMR P = 7.69 × 10−4), chr5:8217236–8217322 (which also tags an intragenic region, DMR P = 1.27 × 10−3), and chr10: 123353418–123355576, which spans a region from the 5’-UTR of FGFR2 to the first exon within FGFR2 (DMR P = 4.79 × 10−3). TABLE 2 Table of DMRs

Differentially methylated regions (DMRs) were called between progressors and nonprogressors via the dmrLasso function of the CHAMP software package (Table 2). Significant DMRs were found from chr2:503065–503193 (which tags an intragenic region, DMR P = 7.69 × 10−4), chr5:8217236–8217322 (which also tags an intragenic region, DMR P = 1.27 × 10−3), and chr10: 123353418–123355576, which spans a region from the 5’-UTR of FGFR2 to the first exon within FGFR2 (DMR P = 4.79 × 10−3). TABLE 2 Table of DMRs DMR ID Probe ID Probe Level Adjusted P Chromosome Gene Start of DMR (bp) End of DMR (bp) Size in bp Change in Methylation P for DMR 1 cg21273584 0.014 2 NA 502999 503195 197 −34% 7.69E-04 1 cg00854591 0.045 2 NA 502999 503195 197 −16% 7.69E-04 1 cg11573608 0.014 2 NA 502999 503195 197 −27% 7.69E-04 2 cg25568703 0.047 5 NA 8216903 8217655 753 −25% 1.27E-03 2 cg25016964 0.039 5 NA 8216903 8217655 753 −22% 1.27E-03 2 cg17642708 0.021 5 NA 8216903 8217655 753 −33% 1.27E-03 3 cg10788901 0.172 10 FGFR2 123352704 123355661 2958 −27% 4.79E-03 3 cg14856220 0.044 10 FGFR2 123352704 123355661 2958 −24% 4.79E-03 3 cg17337672 0.003 10 FGFR2 123352704 123355661 2958 −47% 4.79E-03 3 cg02412684 0.031 10 FGFR2 123352704 123355661 2958 −38% 4.79E-03 3 cg06791446 0.058 10 FGFR2 123352704 123355661 2958 −13% 4.79E-03 3 cg22633036 0.581 10 FGFR2 123352704 123355661 2958 −4% 4.79E-03 We took advantage of the information provided by the 2 color Illumina Infinium chemistry to call copy number aberrations (CANs) within the regions targeted by the methylation probes using the CNA calling function of CHAMP. This did not demonstrate any recurrent copy number alterations between progressors and nonprogressors. There were no significant differences in the numbers of CNA between the 2 groups, with a median of 43 CNAs (range 24–92) in the progressors versus 44 CANs (range 30–52) in the nonprogressors (P = 1.0, Wilcoxon rank-sum).

s did not demonstrate any recurrent copy number alterations between progressors and nonprogressors. There were no significant differences in the numbers of CNA between the 2 groups, with a median of 43 CNAs (range 24–92) in the progressors versus 44 CANs (range 30–52) in the nonprogressors (P = 1.0, Wilcoxon rank-sum). Pathway methylation analysis was carried out using Database for Annotation, Visualisation and Integrated Discovery. Initially Kyoto Encylopedia of Genes & Genomes pathway analysis showed that genes associated with mitogen-activated protein kinase signaling were enriched in the dataset (P = 0.012). Gene ontology analysis using the UP_KEYWORDS feature showed significant enrichment for the disease mutation (P = 9.6 × 10−6), polymorphism (P = 1.7 × 10−5), glycoprotein (P = 3.4 × 10−5), and alternate splicing (P = 1.1 × 10−4) terms. Validation Pyrosequencing Because of the likely biological relevance of FGFR2, and the data demonstrating that OR3A4 was the top differentially methylated CpG, validation pyrosequencing was carried out on all 67 patients. Normality of distribution of methylation values was ascertained by histogram plots, in which it was found that methylation was non-normally distributed; therefore nonparametric testing was carried out.

ng that OR3A4 was the top differentially methylated CpG, validation pyrosequencing was carried out on all 67 patients. Normality of distribution of methylation values was ascertained by histogram plots, in which it was found that methylation was non-normally distributed; therefore nonparametric testing was carried out. For OR3A4 cg09890332, median methylation was 67.8% [interquartile range (IQR) 12.1] in progressors versus 96.7% (IQR 16.1) in nonprogressors (P = 0.0001, z = 5.158; Wilcoxon rank-sum test) (Fig. 2). The pyrosequencing assay design used covered 2 additional CpG +4 bp and +10 bp downstream of cg09890332. Median methylation in these was 66.8% and 59.7% in progressors versus 75.0% and 68.1% in nonprogressors (P = 0.0280 and 0.0368, z = 2.197 and 2.088, Wilcoxon rank-sum test). To investigate whether this phenomenon was localized to this region or was a gene-wide phenomenon, an additional pyrosequencing assay was designed based on probe ID cg07863524 (chr17:3213471–3213471), which is +976 bp downstream from cg09890332 and −68 bp from the transcription start site of OR4A4. This demonstrated that median methylation was 62.2% in progressors and 56.7% in nonprogressors (P = 0.600, z = −0.524; Wilcoxon rank-sum test). A temporal analysis of change in methylation of cg09890332 over time is shown in Fig. 3, showing that the difference between methylation levels at initial biopsy is static between progressors and nonprogressors, and that the difference is maintained over time and is detectable for an extended period of time before diagnosis of EADC.

lysis of change in methylation of cg09890332 over time is shown in Fig. 3, showing that the difference between methylation levels at initial biopsy is static between progressors and nonprogressors, and that the difference is maintained over time and is detectable for an extended period of time before diagnosis of EADC. FIGURE 2 Box plot of OR3A4 methylation differentiating high-risk nondysplastic BE. FIGURE 3 Change in methylation across temporally acquired samples for OR3A4. The y-axis shows percentage methylation at cg09890332 as observed by pyrosequencing. Time-points on the x-axis refer to the sampling points, with 1 representing the initial baseline endoscopy and subsequent visits referred by increasing numbers (surveillance intervals are variable). We then validated cg17337672 within FGFR2, finding that median methylation was 83.4% in progressors versus 82.2% in nonprogressors (P = 0.51, z = 0.653; Wilcoxon rank-sum test).

FIGURE 3 Change in methylation across temporally acquired samples for OR3A4. The y-axis shows percentage methylation at cg09890332 as observed by pyrosequencing. Time-points on the x-axis refer to the sampling points, with 1 representing the initial baseline endoscopy and subsequent visits referred by increasing numbers (surveillance intervals are variable). We then validated cg17337672 within FGFR2, finding that median methylation was 83.4% in progressors versus 82.2% in nonprogressors (P = 0.51, z = 0.653; Wilcoxon rank-sum test). Expression of OR3A4 in Progressors Versus Nonprogressors We then carried out immunohistochemical assessment of expression of OR3A4 (Fig. 4), which, although is labeled as long noncoding RNA, is actually expressed in tissues (see Supplementary results), in a subset of 12 patients. For the stromal compartment, a median expression of 6 (IQR 4–7) was seen in progressors and 2 (IQR 1–3) in nonprogressors (Wilcoxon rank-sum P = 0.0308, z = −2.160). For the epithelial compartment, a median expression of 8 (IQR 8–10) was seen in progressors and 5.5 (IQR 3–10) in nonprogressors (Wilcoxon rank-sum P = 0.4587, z = −0.741). Percentage methylation at OR3A4 and stromal expression was strongly negatively correlated (Pearson correlation coefficient −0.85, P = 0.014), and a similar, but nonsignificant correlation was observed with epithelial expression and methylation at OR3A4 (Pearson correlation coefficient −0.40, P = 0.373).

7, z = −0.741). Percentage methylation at OR3A4 and stromal expression was strongly negatively correlated (Pearson correlation coefficient −0.85, P = 0.014), and a similar, but nonsignificant correlation was observed with epithelial expression and methylation at OR3A4 (Pearson correlation coefficient −0.40, P = 0.373). FIGURE 4 Light micrographs of representative examples of expression of OR3A4 via IHC of nonprogressor (A 10× view, B 40× view) and progressor (C 10× view, D 40×) view. Ability of OR3A4 Methylation to Act as a Discriminator in BE To understand the accuracy of using methylation within cg09890332 of OR3A4 as a biomarker for high-risk BE, we carried out a multivariable reverse stepwise logistic regression analysis of methylation at the 3 tagged CpG dinucleotides within the pyrosequencing assay as the independent variables and progressor versus nonprogressor status as the dependent variables. In this model, CpGs 2 and 3 became nonsignificant (P = 0.3325 and P = 0.4764) and were removed from the model, leaving the first CpG in cg09890332 as being significant [coef = −0.0563, SE = 0.016, z = −3.40, P = 0.001, 95% confidence interval (CI) −0.089 to −0.024]. Using receiver-operator curve modeling, the area under curve of this model was 0.82 (95% CI 0.80–0.83) in the cohort where the marker was originally generated (Supplementary Fig. 2).

CpG in cg09890332 as being significant [coef = −0.0563, SE = 0.016, z = −3.40, P = 0.001, 95% confidence interval (CI) −0.089 to −0.024]. Using receiver-operator curve modeling, the area under curve of this model was 0.82 (95% CI 0.80–0.83) in the cohort where the marker was originally generated (Supplementary Fig. 2). We then used the diagt function of Stata 11.2 to model a set methylation threshold effect on sensitivity and specificity of the test, aiming for maximum negative predictive value and correcting for an incidence rate within the cohort of 0.7%. Modeling at a threshold of below 89% being significant showed that hypomethylation at OR3A4 can predict progression to invasive carcinoma with a sensitivity of 70.8%, specificity of 86%, positive predictive value of 85% and negative predictive value of 72.5%. We then carried out validation bisulfite pyrosequencing on a cohort of progressors (n = 18) versus nonprogressors (n = 14, Southampton cohort), finding that there were significant differences (P = 0.0477, unpaired t test) in methylation, with an average methylation of 59.2% (95% CI 56.2%–62.1%) in progressors versus 63.5% (95% CI 60.2%–66.7%) in the nonprogressors. Regression model demonstrated area under curve 0.70, and adjustment for a prevalence of 0.7% using a threshold of 58% demonstrated a sensitivity of 33.3%, specificity of 78.6%, positive predictive value of 10.5%, and negative predictive value of 94%.

progressors versus 63.5% (95% CI 60.2%–66.7%) in the nonprogressors. Regression model demonstrated area under curve 0.70, and adjustment for a prevalence of 0.7% using a threshold of 58% demonstrated a sensitivity of 33.3%, specificity of 78.6%, positive predictive value of 10.5%, and negative predictive value of 94%. CONCLUSIONS We have identified that hypomethylation at cg09890332 corresponding to the CG nucleotide at position (CHR) of OR3A4 can discriminate between patients who progress from nondysplastic BE and those who did not. This association is maintained across independent cohorts, and seems to be related temporally (ie, the association is maintained in the earliest set of samples from a time series of follow-up biopsies in patients with BE), and also by case status. Gastroesophageal reflux is a key risk factor in the development of BE,19 and also obesity and cigarette smoking.

dependent cohorts, and seems to be related temporally (ie, the association is maintained in the earliest set of samples from a time series of follow-up biopsies in patients with BE), and also by case status. Gastroesophageal reflux is a key risk factor in the development of BE,19 and also obesity and cigarette smoking. The effect of hypomethylation on the OR3A4 gene seems to be functional, in that immunohistochemistry reveals an increase in OR3A4 expression in samples with hypomethylation. The finding that the stromal expression, in particular, is increased is of interest, given the known effect of “pathological” stroma in the pathogenesis of esophageal cancer.37 Our observed region coincides within 225 bp of a CTCF and RAD21 transcription factor binding site, further suggesting that methylation there has a functional effect to prevent transcription factor binding and alter gene expression. Guo et al38 performed a genome-wide screen of long noncoding RNAs in gastric adenocarcinoma, finding that OR3A4 was significantly (55.9-fold) overexpressed in these patients. They also observed that levels of OR3A4 were correlated with metastatic potential and prognosis. Furthermore, they utilized OR3A4 overexpression vectors and performed siRNA knockdown to demonstrate that OR3A4 seems to regulate cellular proliferation in gastric cancer cell lines. Finally, they utilized their overexpressing cell line models and implanted them into nude mice, finding that OR3A4 overexpressing gastric cancer cell lines grew significantly faster and more aggressively than with knockdown of OR3A4. Downstream analysis of target genes demonstrated that OR3A4 targets PDLIM2, a putative tumor suppressor than regulates cell cycle and adhesion; PIWIL1, a transcriptional silencer; and DLX4 which induces epithelial-mesenchymal transition via TWIST1.

ignificantly faster and more aggressively than with knockdown of OR3A4. Downstream analysis of target genes demonstrated that OR3A4 targets PDLIM2, a putative tumor suppressor than regulates cell cycle and adhesion; PIWIL1, a transcriptional silencer; and DLX4 which induces epithelial-mesenchymal transition via TWIST1. We found both at the individual probe level and as part of a DMR that there is hypomethylation in the CpG island associated with FGFR2; however, this did not validate at the single probe level when examined with bisulfite pyrosequencing. FGFR2 has been observed to undergo recurrent alteration in both esophageal adenocarcinoma39 and squamous cell carcinoma,40 with the latter demonstrating recurrent amplification. The disparity between our microarray results and validation by pyrosequencing may be due to probe inflation caused by small sample size, and is a significant weakness of our study; however, given its biological associations with esophageal adenocarcinoma, further work is needed. In common with premalignant lesions in cancer, such as colorectal adenomatous polyps,30 we observed a trend towards genome-wide hypomethylation as demonstrated by a leftward shift of our genome-wide volcano plot, suggesting a widespread overexpression of genes as part of the development towards malignancy. We also found no difference in chromosomal instability between progressors and nonprogressors, although there was widespread instability within both sets of samples, in common with what has previously been observed41 in BE.

, suggesting a widespread overexpression of genes as part of the development towards malignancy. We also found no difference in chromosomal instability between progressors and nonprogressors, although there was widespread instability within both sets of samples, in common with what has previously been observed41 in BE. Another weakness of our study was the inability to carry out a more comprehensive validation of all observed markers as part of a larger panel of markers. Our study made use of extremely small tissue biopsies from endoscopic surveillance programs, which limited the quantity of usable DNA that could be extracted from these samples and used for downstream validation, and thus validation of the observed DMR and other differentially methylated position regions could not be carried out. We were also limited in the number of samples that could be tested, because of the rarity of biopsy samples before the diagnosis of esophageal cancer, as we took advantage of a local screening program to obtain samples. However, in both the genome-wide and in the validation phase, we believe we have sufficient power to detect methylation changes in this marker. In the genome-wide phase, our sample size of 24 patients would allow us to detect a methylation difference42 of 2%, with a statistical power of 90%. Similarly, in the validation phase, our sample size would allow us to detect a minimum methylation change of 10% in the sample set, given the previously observed median methylation and standard deviation in these samples.

tients would allow us to detect a methylation difference42 of 2%, with a statistical power of 90%. Similarly, in the validation phase, our sample size would allow us to detect a minimum methylation change of 10% in the sample set, given the previously observed median methylation and standard deviation in these samples. We observed a median time to diagnosis of esophageal adenocarcinoma on commencement of the surveillance program of 114 months, which we believe is a reflection of the early identification of these patients and their enrolment into a screening program, and the known slow progression of esophageal adenocarcinoma. A further problem with molecular genetic analysis is heterogeneity, due to the low proportion of cells within a biopsy specimen that contain changes compatible with BE, which leads to less clear methylation changes. Hypomethylation of OR3A4, although seemingly accurate for the detection of progression of Barrett to invasive adenocarcinoma, is likely to be of more utility as a multimodal stratifier in BE, taking account of previous findings at the mutational and copy number level, and also epigenetic change. However, for the purpose of designing a surveillance program with the ability to risk-stratify the nondysplastic BE patient, this marker has significant potential utility.

e utility as a multimodal stratifier in BE, taking account of previous findings at the mutational and copy number level, and also epigenetic change. However, for the purpose of designing a surveillance program with the ability to risk-stratify the nondysplastic BE patient, this marker has significant potential utility. Development of a streamlined surveillance program could lead to cost savings through the avoidance of unnecessary upper gastrointestinal endoscopy or via a less invasive technology such as the CytoSponge,43 in a low-risk cohort identified by a molecular marker panel. More frequent endoscopy in the high-risk cohort could lead to earlier diagnosis of EADC or initiate management of BE to arrest further progression. In conclusion, development of a stratified marker panel in the context of a clinical trial is now needed to improve diagnosis of high-risk BE. Supplementary Material Supplemental Digital Content MPD, TN, OT, and ADB contributed equally to this study. Funding: ADB acknowledges funding from the Wellcome Trust (102732/Z/13/Z), Cancer Research UK (C31641/A23923) and the Medical Research Council (MR/M016587/1). ADB also acknowledges support from the University of Birmingham Human Biomaterials Resource Centre. MPD and TN acknowledge funding from the QE Hospital Charities. TJU acknowledges funding by the Medical Research Council (G1002565) and Cancer Research UK (C10104/A23924). TJU also acknowledges the University of Southampton CRUK/ECMC Tissue Bank. The authors declare no conflict of interests.

Parastomal hernia (PSH) is a common complication after stomal surgery. Though the exact incidence has not been fully established, figures range between a few and 78%.1–3 Most PSHs develop within 2 years of surgery but can occur up to 30 years later. Approximately, one third of patients operated for rectal cancer in Sweden have a permanent colostomy, and 24% to 39% of patients with a “temporary” stoma never have it reversed.4–6 PSH causes difficulties with stoma dressing and leakage, increases the risk for incarceration, and has a negative impact on the patient's quality of life.7–9 A number of strategies have been proposed to prevent the formation of PSH after primary surgery: choice of stoma placement through versus lateral to the rectus sheath, transperitoneal versus extraperitoneal, and correct sizing of the trephine.2,10 None of these seems to reduce the incidence of PSH. Furthermore reported 30-day morbidity and mortality rates of planned repair procedures are 8% to 36% and 0% to 5%, respectively 11–13 Emergency PSH repair has a reported mortality rate of 11% to 25%.14,15 The development of stoma techniques that reduce the risk for PSH is thus a field of research that should be given priority.

ore reported 30-day morbidity and mortality rates of planned repair procedures are 8% to 36% and 0% to 5%, respectively 11–13 Emergency PSH repair has a reported mortality rate of 11% to 25%.14,15 The development of stoma techniques that reduce the risk for PSH is thus a field of research that should be given priority. Placement of a mesh to prevent PSH formation has been proposed. Eight small and 2 large randomized controlled trials (RCT) have been published during the past 2 decades. Pooled data from these studies show promising results,16 though mesh location, choice of mesh material and surgical approach varied.17–23 The use of mesh did not increase complication rate, but none of the trials was designed to assess complication. On the contrary, 1 retrospective trial comparing PSH rate before and after the introduction of routine prophylactic mesh around the stoma, at a unit that repeatedly produced top results in the Swedish Rectal Cancer Registry, showed no difference in complication rates between groups.24 The Swedish National Board of Health and Welfare classifies the use of prophylactic mesh around stomas as, “Research and Development,” because of the weak scientific evidence; larger RCTs comparing prophylactic mesh with no mesh are required.25

ncer Registry, showed no difference in complication rates between groups.24 The Swedish National Board of Health and Welfare classifies the use of prophylactic mesh around stomas as, “Research and Development,” because of the weak scientific evidence; larger RCTs comparing prophylactic mesh with no mesh are required.25 In view of this uncertainty, we designed a double-blinded multicenter RCT to evaluate the effects of using a prophylactic mesh around the stoma. The primary aim was to evaluate the PSH rate in both groups, and a secondary aim was to assess the risk for early complications. Our hypothesis was that polypropylene mesh in the sublay position around a colostomy decreases the risk for PSH. METHODS Study Design The study was designed as a double-blinded multicenter RCT. Patients were randomized to 2 groups, those with and those without prophylactic mesh around the stoma. The study protocol adhered to the Helsinki Declaration and was approved by the Regional Ethics Committee at Umeå University, Sweden (DNR 07-081 M). The study was registered at ClinicalTrials.gov (Identifier: NCT00917995) and adhered to the CONSORT 2010 criteria for RCTs. Patients All patients scheduled for permanent colostomy at one of the participating hospitals and who met the inclusion criteria (no previous stoma; older than 18 years; and with informed consent), were eligible to participate in the study. Exclusion criteria were: expected survival less than 3 years; fecal peritonitis; previous stoma; and no informed consent.

ent colostomy at one of the participating hospitals and who met the inclusion criteria (no previous stoma; older than 18 years; and with informed consent), were eligible to participate in the study. Exclusion criteria were: expected survival less than 3 years; fecal peritonitis; previous stoma; and no informed consent. Surgery Colostomy without mesh was taken through the rectus muscle, at a site marked by a stoma therapist before surgery. The height of the stomal nipple was at least 1 cm, and neither intraabdominal lateral closure nor fixation to the fascia was performed. Mucocutaneous interrupted monofilament sutures were used to fix the stoma to the skin.

t mesh was taken through the rectus muscle, at a site marked by a stoma therapist before surgery. The height of the stomal nipple was at least 1 cm, and neither intraabdominal lateral closure nor fixation to the fascia was performed. Mucocutaneous interrupted monofilament sutures were used to fix the stoma to the skin. Colostomy with mesh incorporated a lightweight polypropylene mesh (density 25–40 g/m2) in the sublay position. A 10 × 10 cm space for the mesh was created dorsal to the rectus abdominis muscle but anterior to the posterior rectus sheath and peritoneum. The bowel was passed through the rectus muscle via a cruciform incision in the middle of the mesh. Single nonabsorbable monofilament sutures anchored the mesh laterally in the pocket. Medially the mesh was incorporated in the running suture closing the fascia. A potential risk is that the mesh would not reach the midline; however, no such effect was described in the original article.26 The surgical technique used had been described in an earlier study comparing stoma with and without prophylactic mesh.26 An instruction video, made by Israelsson et al at the Department of Surgery, Sundsvall, was distributed to all participating centers to ensure that a uniform surgical technique was used. Fixation to the skin was performed with the same technique as when creating a colostomy without mesh. Surgery was performed by an experienced colorectal surgeon with an annual volume of at least 100 major surgical procedures. Postoperative mobilization was according to each hospital's routine.

Colostomy with mesh incorporated a lightweight polypropylene mesh (density 25–40 g/m2) in the sublay position. A 10 × 10 cm space for the mesh was created dorsal to the rectus abdominis muscle but anterior to the posterior rectus sheath and peritoneum. The bowel was passed through the rectus muscle via a cruciform incision in the middle of the mesh. Single nonabsorbable monofilament sutures anchored the mesh laterally in the pocket. Medially the mesh was incorporated in the running suture closing the fascia. A potential risk is that the mesh would not reach the midline; however, no such effect was described in the original article.26 The surgical technique used had been described in an earlier study comparing stoma with and without prophylactic mesh.26 An instruction video, made by Israelsson et al at the Department of Surgery, Sundsvall, was distributed to all participating centers to ensure that a uniform surgical technique was used. Fixation to the skin was performed with the same technique as when creating a colostomy without mesh. Surgery was performed by an experienced colorectal surgeon with an annual volume of at least 100 major surgical procedures. Postoperative mobilization was according to each hospital's routine. Study Outcome Early complications were evaluated at 1 month, and late complications and possible recurrence of parastomal hernia were assessed at a 1-year follow up. The examiner was blinded to the group the patient was randomized to. Primary endpoint was the rate of parastomal hernia in both groups, judged clinically and by computerized tomography (CT) 1 year after surgery. Clinical examination was performed in both upright and supine positions by digital palpation within the stoma at rest and during the Valsalva maneuver. Findings were classified as a bulge or PSH. Definition of a bulge was a protrusion around the stoma noticed by the surgeon and/or the patient but deemed not to be a hernia clinically. CT was performed in the supine position, with or without intravenous contrast; the Valsalva maneuver was not performed. All CT examinations were evaluated by a radiologist with a special interest in parastomal hernia. The findings were classified according to the Moreno-Matias scale, classified as validated by the European Hernia Society, based on the content of the hernia: no hernia = 0; bowel forming the stoma with hernia sac < 5 cm = IA; bowel forming the stoma with hernia sac > 5 cm = IB; hernia sac containing omentum = II; and hernia sac containing loop other than bowel forming the stoma = III.27,28

fied as validated by the European Hernia Society, based on the content of the hernia: no hernia = 0; bowel forming the stoma with hernia sac < 5 cm = IA; bowel forming the stoma with hernia sac > 5 cm = IB; hernia sac containing omentum = II; and hernia sac containing loop other than bowel forming the stoma = III.27,28 Sample Size As the vast majority of patients receiving a colostomy are operated for colorectal cancer, a 5-year survival rate around 50% is likely. Assuming a parastomal hernia incidence of 20% 7 without and 5% with mesh, a 5-year survival rate of just over 50%, and a significance level of 95%, it would require 220 patients to achieve a power of 80%. Randomization and Blinding Randomization was performed using sealed envelopes, stratified per hospital in blocks of four, to ensure balance between the 2 arms. The envelopes were prepared by the Regional Cancer Center North, Umeå University Hospital, in bundles of 100 per hospital, without the help of the investigators. Neither the patient nor the surgeon assessing the patient postoperatively was informed on which arm the patient had been randomized to. Postoperative assessment was made by a surgeon not involved in the primary procedure. If randomization of a patient was performed incorrectly, the patient was replaced by 3 new patients to ensure maintenance of power. Data Collection Data were collected prospectively and patients were followed by protocol with individual case report forms (CRF). All data were entered into an Excel-database.

Randomization and Blinding Randomization was performed using sealed envelopes, stratified per hospital in blocks of four, to ensure balance between the 2 arms. The envelopes were prepared by the Regional Cancer Center North, Umeå University Hospital, in bundles of 100 per hospital, without the help of the investigators. Neither the patient nor the surgeon assessing the patient postoperatively was informed on which arm the patient had been randomized to. Postoperative assessment was made by a surgeon not involved in the primary procedure. If randomization of a patient was performed incorrectly, the patient was replaced by 3 new patients to ensure maintenance of power. Data Collection Data were collected prospectively and patients were followed by protocol with individual case report forms (CRF). All data were entered into an Excel-database. After completion of inclusion, a separate analysis was performed to describe the cohort of patients who met the inclusion criteria but were not included in the study. To obtain a representative group of “not included” patients, all other colostomy operations performed during the inclusion/trial period at Umeå University Hospital, Sunderby Regional Hospital and Mora County Hospital were assessed for demographic data and reasons for not being included.

were not included in the study. To obtain a representative group of “not included” patients, all other colostomy operations performed during the inclusion/trial period at Umeå University Hospital, Sunderby Regional Hospital and Mora County Hospital were assessed for demographic data and reasons for not being included. Statistical Analysis The IBM SPSS Statistics 24 software program was used. All analyses of group variables were performed with the χ2 test and all analyses of continuous variables were performed with the independent Student t test; a P< 0.05 was considered significant. The population was tested for skewness and kurtosis (measure describing outliers in a data set) using SPSS Statistics. As normal distribution was shown, parametric statistics were used. RESULTS Eight hospitals participated in the study: 4 university hospitals (Karolinska University Hospitals in Huddinge and Solna, Stockholm, Uppsala University Hospital, and Umeå University hospital); 2 regional hospitals (Sunderby Regional Hospital and Helsingborg Regional Hospital); and 2 county hospitals (Mora County Hospital and Nyköping County Hospital). The total catchment area of the 8 hospitals is approximately 1,475,000 individuals; approximately 15% of the Swedish population.

and Umeå University hospital); 2 regional hospitals (Sunderby Regional Hospital and Helsingborg Regional Hospital); and 2 county hospitals (Mora County Hospital and Nyköping County Hospital). The total catchment area of the 8 hospitals is approximately 1,475,000 individuals; approximately 15% of the Swedish population. From December 2007 through October 2015, 240 patients were enrolled in the trial (Fig. 1). All operations were planned and performed according to a standard protocol within the National Health Service. Eight patients were enrolled in the study but had to be excluded because of change in surgical approach during the operation. Seven of these patients had been excluded and 21 new patients included and randomized according to the study design. Unfortunately, when all CRFs had reached the study center, one more patient had to be excluded, and could not be replaced as the trial had been closed. FIGURE 1 Consort diagram. There was no significant difference in baseline data between the mesh group and the nonmesh group (Table 1) apart from duration of surgery, which was significantly longer in the mesh group (P = 0.019). TABLE 1 Basic Demographic Data.

From December 2007 through October 2015, 240 patients were enrolled in the trial (Fig. 1). All operations were planned and performed according to a standard protocol within the National Health Service. Eight patients were enrolled in the study but had to be excluded because of change in surgical approach during the operation. Seven of these patients had been excluded and 21 new patients included and randomized according to the study design. Unfortunately, when all CRFs had reached the study center, one more patient had to be excluded, and could not be replaced as the trial had been closed. FIGURE 1 Consort diagram. There was no significant difference in baseline data between the mesh group and the nonmesh group (Table 1) apart from duration of surgery, which was significantly longer in the mesh group (P = 0.019). TABLE 1 Basic Demographic Data. Nonmesh Group n = 118 Mesh Group n = 114 P Age, yrs, range 69.9 (35–89) 69.7 (41–86) 0.877 Sex: male 62 (52%) 74 (65%) 0.056 ASA 1+2* 78 (70%) 82 (74%) 0.578 ASA 3* 34 (29%) 28 (26%) 0.578 Smoker† 8 (7%) 12 (11%) 0.326 BMI, range 26.3 (18.5–43.7) 26.1 (16.7–37.8) 0.766 Cancer 106 (90%) 106 (93%) 0.392 Operation time, mins, range 287 (84–625) 323 (70–616) 0.019 Planned surgery 118 (100%) 113 (99%) 0.308 Length of stay, d, range 12 (2–44) 12 (4–44) 0.792 Basic Demographic Data in the Trial Population. *ASA-class was stated in 222 patients. †Smoking habit was stated in 216 patients. Median values are shown for: age, BMI, operation time and length of stay.

Nonmesh Group n = 118 Mesh Group n = 114 P Age, yrs, range 69.9 (35–89) 69.7 (41–86) 0.877 Sex: male 62 (52%) 74 (65%) 0.056 ASA 1+2* 78 (70%) 82 (74%) 0.578 ASA 3* 34 (29%) 28 (26%) 0.578 Smoker† 8 (7%) 12 (11%) 0.326 BMI, range 26.3 (18.5–43.7) 26.1 (16.7–37.8) 0.766 Cancer 106 (90%) 106 (93%) 0.392 Operation time, mins, range 287 (84–625) 323 (70–616) 0.019 Planned surgery 118 (100%) 113 (99%) 0.308 Length of stay, d, range 12 (2–44) 12 (4–44) 0.792 Basic Demographic Data in the Trial Population. *ASA-class was stated in 222 patients. †Smoking habit was stated in 216 patients. Median values are shown for: age, BMI, operation time and length of stay. Comparison of patients in the “not included” group, that is, meeting the inclusion criteria but not included, with those in the STOMAMESH trial group, showed significantly lower proportions undergoing emergency surgery (P < 0.001) and with ASA-class 3 (P < 0.001) in the study group. Furthermore, patients in the study population were older (P = 0.010) (Table 2). TABLE 2 Study population vs “Not included”. STOMAMESH Population n = 232 Not Included n = 434 P Age, yrs, range 69.8 67.5 0.010 ASA 1+2 167 (71.9%) 229 (52.8%) <0.001 ASA 3 65 (28.1%) 205 (47.2%) <0.001 Sex: male 135 (58.0%) 217 (50%) 0.020 Planned surgery 231 (99.6%) 337 (77.6%) <0.001 The group of patients included in the STOMAMESH trial compared with the patients who could have been but were “Not included” at 3 hospitals (Sunderby hospital, Umeå University hospital, and Mora hospital) during the inclusion period. Median values are shown for age.

0.020 Planned surgery 231 (99.6%) 337 (77.6%) <0.001 The group of patients included in the STOMAMESH trial compared with the patients who could have been but were “Not included” at 3 hospitals (Sunderby hospital, Umeå University hospital, and Mora hospital) during the inclusion period. Median values are shown for age. When excluding emergency procedures from both groups differences in age, ASA, and operation time between the 2 groups remained (Table 3). TABLE 3 Study population vs "Not included" emergency cases excluded. STOMAMESH Population n = 231 Not Included n = 337 P Age, yrs, range 69.8 67.7 0.043 ASA 1+2 166 (71.9%) 200 (59.3%) 0.002 ASA 3 65 (28.1%) 137 (40.7%) 0.002 Sex: male 135 (58.0%) 164 (48.7%) 0.029 The group of patients included in the STOMAMESH trial compared with the patients who could have been but were “Not included” at 3 hospitals (Sunderby hospital, Umeå University hospital, and Mora hospital) during the inclusion period. Emergency procedures and ASA 4 excluded. Median values are shown for age. No significant differences were found when data were stratified according to hospital.

STOMAMESH Population n = 231 Not Included n = 337 P Age, yrs, range 69.8 67.7 0.043 ASA 1+2 166 (71.9%) 200 (59.3%) 0.002 ASA 3 65 (28.1%) 137 (40.7%) 0.002 Sex: male 135 (58.0%) 164 (48.7%) 0.029 The group of patients included in the STOMAMESH trial compared with the patients who could have been but were “Not included” at 3 hospitals (Sunderby hospital, Umeå University hospital, and Mora hospital) during the inclusion period. Emergency procedures and ASA 4 excluded. Median values are shown for age. No significant differences were found when data were stratified according to hospital. After 1 year, 211 of 232 patients included (91%) were followed up clinically, and 198 of 232 patients (85%) were examined with CT (Fig. 1). According to the requisites for the original power calculation, this gives a de facto power of 91%. Reasons for being lost to follow up were: incomplete follow up because of surgical complication (3 patients); progression of disease (1 patient); development of dementia (1 patient); refusal of patient to participate further in the study (9 patients); and death (7 patients). When comparing the mesh and nonmesh groups, there was no significant difference in the rate of bulge (P = 0.631), clinically judged PSH (P = 0.866), or PSH on the CT (P = 0.748) (Table 4). There was no difference in complication rates at 1-month follow up between the groups (Table 5). Reoperations performed during the first postoperative month were because of small bowel obstruction requiring adhesiolysis (1 in each group), conversion to transverse colostomy (1 in the mesh group), revision of the enterostomy (2 in the mesh group and 3 in the nonmesh group), bleeding (2 in the mesh group), superficial/deep infection (1 in each group), and wound rupture (1 in the nonmesh group). A further 2 patients in the nonmesh group and 4 patients in the nonmesh group were reoperated within the first postoperative year. At the 1-year follow up, a total of 8 patients in the nonmesh and 12 in the mesh group had been reoperated.

ial/deep infection (1 in each group), and wound rupture (1 in the nonmesh group). A further 2 patients in the nonmesh group and 4 patients in the nonmesh group were reoperated within the first postoperative year. At the 1-year follow up, a total of 8 patients in the nonmesh and 12 in the mesh group had been reoperated. TABLE 4 Hernia rate at one year. Nonmesh Group n = 107 Mesh Group n = 104 P Hernia (judged clinically) 32 (30%) 30 (29%) 0.866 Bulge, no hernia (judged clinically) 18 (17%) 15 (14%) 0.631 Hernia classified 2 and 3 on CT 28 (26%) 25 (24%) 0.748 Hernia classified 1, 2, and 3 on CT 36 (34%) 33 (32%) 0.765 Follow Up at 1 Year Evaluating Presence of Parastomal Hernia. CT-scan was performed on 99 patients in each group and the findings were classified according to the model by Moreno-Matias. Bulge was defined as a protrusion around the stoma judged not to be a hernia. TABLE 5 Complications at one month. Nonmesh Group n = 112 Mesh Group n = 110 P Overall complications 36 (32%) 38 (34%) 0.668 Surgical complications 31 (28%) 30 (27%) 0.947 Wound infection 16 (14%) 17 (15%) 0.785 Deep infection 9 (8%) 7 (6%) 0.644 Intestinal obstruction 1 (1%) 2 (2%) 0.618* Stoma necrosis 8 (7%) 5 (4%) 0.570* Reoperation within 30 days 6 (5%) 7 (6%) 0.783* Other complications Acute myocardial infarction 2 (2%) 2 (2%) 1.000* Pneumonia 1 (1%) 2 (2%) 0.618* Thrombosis 0 2 (2%) 0.242* Urinary tract infection 1 (1%) 5 (4%) 0.116* Complications at the 1-Month Follow Up in the Trial Population. *The assumption for the χ2 test was not fulfilled in these comparisons and Fischer exact test was used.

Nonmesh Group n = 112 Mesh Group n = 110 P Overall complications 36 (32%) 38 (34%) 0.668 Surgical complications 31 (28%) 30 (27%) 0.947 Wound infection 16 (14%) 17 (15%) 0.785 Deep infection 9 (8%) 7 (6%) 0.644 Intestinal obstruction 1 (1%) 2 (2%) 0.618* Stoma necrosis 8 (7%) 5 (4%) 0.570* Reoperation within 30 days 6 (5%) 7 (6%) 0.783* Other complications Acute myocardial infarction 2 (2%) 2 (2%) 1.000* Pneumonia 1 (1%) 2 (2%) 0.618* Thrombosis 0 2 (2%) 0.242* Urinary tract infection 1 (1%) 5 (4%) 0.116* Complications at the 1-Month Follow Up in the Trial Population. *The assumption for the χ2 test was not fulfilled in these comparisons and Fischer exact test was used. DISCUSSION This trial, the largest multicenter RCT on stomal surgery in routine healthcare, revealed no difference in PSH rate between procedures incorporating a prophylactic mesh around the stoma and those with no mesh. Complication and reoperation rates did not differ between groups. Compared with previously published trials analyzing the use of mesh in stomal surgery, the current trial was based on a larger population and a systematic follow up including both clinical and CT assessment of PSH. Furthermore, this study was multicenter and thus represents the use of mesh in normal clinical practice using well-established routines. It covers most regions in Sweden, and represents approximately 15% of the population. Furthermore, unlike other studies, this study was blinded to both patient and examiner, which could be an explanation for the diverging results.18–20,26 This trial also includes different types of hospital and levels of specialization, whereas most previous trials are from single specialist Institutions.20,22,26,29 The trial reached a de facto power of 91%. The only significant difference found between the 2 treatment arms was a longer operation time (median 38 minutes) for the mesh group corresponding to the time expected for application of the mesh. Interestingly, in a large retrospective study in Sweden including 206 patients, reported from a single institution, it was not possible to show a difference in PSH between procedures with or without mesh.24 Follow-up assessment in that trial also included both clinical investigation and CT. That trial was carried out at a high-volume single center with top results in the Swedish Rectal Cancer Registry.4 This center did not participate in the STOMAMESH trial, but had similar results.

ween procedures with or without mesh.24 Follow-up assessment in that trial also included both clinical investigation and CT. That trial was carried out at a high-volume single center with top results in the Swedish Rectal Cancer Registry.4 This center did not participate in the STOMAMESH trial, but had similar results. We also describe the patient group eligible for inclusion but not taking part, the “not included” group, a procedure rarely performed in previous trials.18,19,21,22,26,29 A potential weakness of this study is that we choose to investigate the population of “not included” patients at 3 hospitals (Sunderby, Umeå, and Mora), to get a representative sample, instead of doing this at all participating hospitals. Furthermore, the choice to use routine CT might also be considered a weakness. In some studies, the patients were placed in the supine position or performed the Valsalva maneuver during the examination.30 This might have increased the sensitivity of CT in this study. The reason not doing this is that abdominal CT scan with the patient in the supine position is standard. An additional CT with Valsalva would have increased radiation exposure and would have demanded extra resources, thus deviating from routine practice. The majority of CT scans were included in routine colorectal cancer follow up at the hospitals participating. It would seem unlikely; however, that this difference in CT follow up should favor one or other of the treatment arms.

ion exposure and would have demanded extra resources, thus deviating from routine practice. The majority of CT scans were included in routine colorectal cancer follow up at the hospitals participating. It would seem unlikely; however, that this difference in CT follow up should favor one or other of the treatment arms. The reason for choosing the sublay polypropylene mesh technique was that this had been investigated in a number of previous studies and that laparoscopic surgery had not been accepted as standard practice in Sweden, 2005. Open surgery was thus the natural choice. Only 1 patient in the STOMAMESH population had emergency surgery whereas 96 (22.4%) in the “not included” population were operated as an emergency. This was a systematic error in the study design but does not entirely explain the differences in ASA and age. One might argue that we should have excluded emergency procedures in the study design. Consequently, we cannot extrapolate the results from elective surgery to those having emergency surgery with a contaminated abdomen.

a systematic error in the study design but does not entirely explain the differences in ASA and age. One might argue that we should have excluded emergency procedures in the study design. Consequently, we cannot extrapolate the results from elective surgery to those having emergency surgery with a contaminated abdomen. This study did not investigate or quantify symptoms arising from PSH. It is well known that not all PSHs actually cause symptoms 7 and this has been used as an argument against the use of prophylactic mesh. The PSH rate in our study was more than three times that in a recent review article (29.4% vs 8.9%),16 implying that results from smaller studies performed at single institutions with a special interest in PSH are difficult to extrapolate to routine clinical practice. The meticulous follow up in this study might well have generated the higher PSH rates. No negative side-effects of prophylactic mesh were detected despite the rather large trial population.

studies performed at single institutions with a special interest in PSH are difficult to extrapolate to routine clinical practice. The meticulous follow up in this study might well have generated the higher PSH rates. No negative side-effects of prophylactic mesh were detected despite the rather large trial population. There have been several trials studying PSH, and the majority has shown positive results regardless of which prophylactic mesh is used when forming a colostomy.16 These authors recommend the use of mesh when creating a colostomy. The results from the current study do not support this conclusion. Comparing our trial with the Dutch trial,18 which was second largest and the one most similar in design to ours, the Dutch trial did not include routine CT follow up and blinding which could explain the difference in results. In view of the dubious need for prophylactic mesh and the divergent trial results to date, we cannot recommend the use of prophylactic mesh when creating a colostomy. Furthermore, we do not have a clear picture of how many patients who develop a PSH suffer symptoms related to their PSH. The most promising results have been reported using mesh for repair of PSH, including the Sugarbaker procedure.29,31 Morbidity risk with mesh repair is high and deaths have been reported.16,32,33 There are reports of long-term complications from mesh implantation in the form of discomfort, pain, and fistula formation.33–35 Should mesh be used prophylactically on a large scale, the incidence of these complications would likely increase. In fact, long-term follow-up studies on the use of prophylactic parastomal mesh are rare indeed.

long-term complications from mesh implantation in the form of discomfort, pain, and fistula formation.33–35 Should mesh be used prophylactically on a large scale, the incidence of these complications would likely increase. In fact, long-term follow-up studies on the use of prophylactic parastomal mesh are rare indeed. CONCLUSIONS In routine surgical care, no difference in rate of PSH can be seen when creating an elective colostomy with or without a reinforcing mesh. The management of PSH is still a considerable surgical challenge and there is no generally accepted method for its prevention. Further research should focus on developing preventive measures and improving methods for PSH repair. Acknowledgments The authors would like to thank research nurse Lisa Eskilsson, Department of Surgery, Sunderby hospital, Region Norrbotten; Robert Lundqvist, statistician, Region Norrbotten; and all hospitals participating in the study. Disclosure: Financial support was received from the Swedish Research Council 214–7196 and the Department of Research and Development, Region Norrbotten. The authors report no conflicts of interest.

Colorectal cancer (CRC) is a common disease in the UK,1 and prognosis is poor if, at diagnosis, the disease has metastasized. Unfortunately, this is all too common with approximately 20% of the 42,000 people diagnosed annually1 having metastatic disease at presentation2–4 and up to 50% subsequently going on to develop it during the course of their illness.5 UK outcomes from CRC are known to lag behind many economically comparable countries6 and as this may, at least in part, be because of poorer outcomes for individuals who present with late stage disease, optimizing their care is a priority. In England, of the 35,000 people diagnosed with CRC, <60% undergo surgical resection of the primary tumor: about 21,000 cases per annum.

ically comparable countries6 and as this may, at least in part, be because of poorer outcomes for individuals who present with late stage disease, optimizing their care is a priority. In England, of the 35,000 people diagnosed with CRC, <60% undergo surgical resection of the primary tumor: about 21,000 cases per annum. It is estimated that at least 15% to 20% of patients with metastases in the liver may be eligible for potentially curative liver resection.7–9 The National Institute of Health and Clinical Excellence guidelines recommend liver resection as the treatment of choice for metastatic disease in patients where this is possible and 10-year survival following such operations is almost 25%.10 Unfortunately, previous studies11–13 have shown that the rates of resection for CRC liver metastases (CRCLM) vary significantly across the English National Health Service (NHS) indicating there may be considerable variation in the decision-making process as to who is, and is not, eligible for resection and who is referred to a specialist liver team.12 The majority of the available evidence dates from >10 years ago so more contemporary data are urgently required to address the persistent concern that there may still be some patients being denied access to a specialist liver services and potentially curative treatments.13

who is referred to a specialist liver team.12 The majority of the available evidence dates from >10 years ago so more contemporary data are urgently required to address the persistent concern that there may still be some patients being denied access to a specialist liver services and potentially curative treatments.13 This population-based study aims, therefore, to provide this information by investigating the frequency of surgical resections for CRCLM across the English NHS using the most recently available data. Trends are examined in relation to patient and tumor characteristics to identify whether the trends in likelihood of resection in relation to these factors and hospital of treatment still persist.

investigating the frequency of surgical resections for CRCLM across the English NHS using the most recently available data. Trends are examined in relation to patient and tumor characteristics to identify whether the trends in likelihood of resection in relation to these factors and hospital of treatment still persist. METHODS Study Population and Data Definitions All adults diagnosed with a first primary CRC [International Classification of Disease (ICD)-10 codes C18-C20], and who had undergone a major resection for their disease in an NHS hospital with a CRC multidisciplinary team (MDT) between January 1, 2005 and December 31, 2012 (to allow 3 years of follow-up until censoring at December 2015), were identified in the COloRECTal cancer data Repository (CORECT-R). This population-based resource14 contains numerous linked datasets relevant to CRC. For this study, information was derived from a linked National Cancer Registration and Analysis Service and Hospital Episode Statistics (HES) dataset. Information on date of diagnosis, age, sex, deprivation [measured via the income domain of the Index of Multiple Deprivation (IMD) 2010], site of tumor, and stage were extracted from the cancer registry dataset. Where patients had >1 tumor recorded simultaneously, the tumor with the highest stage was selected. Any remaining duplicate patient records were cleaned to select the most relevant tumor for the type of major resection carried out. Information on the type and date of the first major resection surgery following diagnosis was extracted from the HES component of CORECT-R. Major primary resection and liver resection were identified by the appropriate OPCS4.8 codes (Table 1). Primary tumors in the cecum, appendix, ascending colon, hepatic flexure, and transverse colon (ICD-10 codes C18.0-C18.4) were assigned as right-sided colon tumors, whereas tumors in the splenic flexure, descending colon and sigmoid colon (ICD-10 C18.5-C18.7) were assigned as left-sided colon tumors. Tumors in the rectosigmoid (ICD-10 C19) were classified separately because of different characteristics, as were rectal tumors (ICD-10 C20).

were assigned as right-sided colon tumors, whereas tumors in the splenic flexure, descending colon and sigmoid colon (ICD-10 C18.5-C18.7) were assigned as left-sided colon tumors. Tumors in the rectosigmoid (ICD-10 C19) were classified separately because of different characteristics, as were rectal tumors (ICD-10 C20). A Charlson comorbidity index15 was derived for each individual in the cohort, taking into account diagnoses (excluding cancer) from any hospital admissions in the year before CRC diagnosis. The cancer component of the Charlson index was derived from the cancer registry information found in CORECT-R and the score for any other cancers in the year before CRC was added to that obtained from HES data. The Charlson score was categorized as: 0, 1, 2, and ≥3 with higher scores indicating greater comorbidity.

is. The cancer component of the Charlson index was derived from the cancer registry information found in CORECT-R and the score for any other cancers in the year before CRC was added to that obtained from HES data. The Charlson score was categorized as: 0, 1, 2, and ≥3 with higher scores indicating greater comorbidity. In the English NHS, a Trust is an organization, comprising ≥1 hospitals that provide care to patients in a city or region. In this article, we have used the term “hospital” in place of Trust. Data regarding whether the initial colorectal resection was carried out within a hospital with an onsite hepatobiliary team (HBT) were obtained from the Organizational Survey 2016, carried out by the National Bowel Cancer Audit (NBOCA).16 As the survey is a snapshot, carried out at a date outside of the range of this dataset, the location of HBTs may change with service redesign. The survey results were checked against activity in the HES data. Liver center status was retained for hospitals where >80% of liver resections were carried out in the same hospital as patients’ colorectal primary resection, and where hepatic resection activity was consistent across the whole study period. This provided good agreement with the Organizational Survey. Primary Endpoints The rate of liver resection was compared across NHS hospitals, to ascertain whether there is significant variation in the surgical management of CRC liver metastases between different providers. Analyses were undertaken with and without adjustment for casemix.

In the English NHS, a Trust is an organization, comprising ≥1 hospitals that provide care to patients in a city or region. In this article, we have used the term “hospital” in place of Trust. Data regarding whether the initial colorectal resection was carried out within a hospital with an onsite hepatobiliary team (HBT) were obtained from the Organizational Survey 2016, carried out by the National Bowel Cancer Audit (NBOCA).16 As the survey is a snapshot, carried out at a date outside of the range of this dataset, the location of HBTs may change with service redesign. The survey results were checked against activity in the HES data. Liver center status was retained for hospitals where >80% of liver resections were carried out in the same hospital as patients’ colorectal primary resection, and where hepatic resection activity was consistent across the whole study period. This provided good agreement with the Organizational Survey. Primary Endpoints The rate of liver resection was compared across NHS hospitals, to ascertain whether there is significant variation in the surgical management of CRC liver metastases between different providers. Analyses were undertaken with and without adjustment for casemix. Statistical Methods The frequency of liver resections was assessed in relation to the year of CRC resection, patient age, sex, site and stage of the primary tumor at diagnosis, IMD quintile, Charlson score, region (based on the Cancer Alliance or Vanguard that the hospital resides in), and the hospital where the initial colorectal resection took place. The statistical significance of any differences in liver resection rates across patient characteristics and between hospitals were assessed using the chi-square test.

core, region (based on the Cancer Alliance or Vanguard that the hospital resides in), and the hospital where the initial colorectal resection took place. The statistical significance of any differences in liver resection rates across patient characteristics and between hospitals were assessed using the chi-square test. Multilevel binary logistic regression was used to determine factors associated with the use of resection for CRCLM, with a hierarchy of patients (level 1) clustered within hospitals (level 2), clustered further into regional Cancer Alliances (level 3). Explanatory variables in the risk-adjusted model were age at resection, sex, IMD quintile, primary tumor site, year of primary major colorectal resection, Charlson comorbidity score, stage at diagnosis, and whether the hospital was a liver center. Regression analysis was also repeated for those diagnosed with synchronous metastatic disease (Stage IV at diagnosis) and those who developed metachronous metastatic disease (Stages I-III at diagnosis). Funnel plots were constructed to show the variation across hospitals and Cancer Alliances using the Spiegelhalter approach17 and those hospitals or Alliances outside of the 99.8% control limits were considered outliers. RESULTS Study Population During the period January 1, 2005 to December 31, 2012, 157,383 patients were identified as undergoing major resection for a colorectal tumor, the characteristics of which are outlined in Table 2. Of these 7423 (4.7%) underwent ≥1 liver resections within 3 years of their primary colorectal resection.

S Study Population During the period January 1, 2005 to December 31, 2012, 157,383 patients were identified as undergoing major resection for a colorectal tumor, the characteristics of which are outlined in Table 2. Of these 7423 (4.7%) underwent ≥1 liver resections within 3 years of their primary colorectal resection. There were 117 patients who had 2 liver resections within 90 days of each other, suggesting a planned 2-stage liver resection. These are therefore treated as 1 resection episode. There were 750 patients who suffered disease recurrence requiring a further resection within the 3-year period. Overall, there were 6673 patients (89.9%) having 1 resection episode, 692 (9.3%) having 2 resection episodes, and 58 (0.8%) having ≥3, giving a total of 8234 liver resection episodes. Most patients (6708, 90.4%) received their first liver resection after their primary colorectal resection, 499 (6.7%) had a synchronous liver resection with their primary colorectal resection and a small minority (216, 2.9%) had their first liver resection carried out before their major colorectal resection. Overall, 2236 (18.9%) of patients who presented with synchronous metastases, received a liver resection within 3 years of their primary colorectal resection. Median time from primary colorectal resection to first liver resection was 270 days.

ir first liver resection carried out before their major colorectal resection. Overall, 2236 (18.9%) of patients who presented with synchronous metastases, received a liver resection within 3 years of their primary colorectal resection. Median time from primary colorectal resection to first liver resection was 270 days. Variation in Resection Rates Over Time The percentage of patients receiving liver resection increased between 2005 and 2012 and appeared to have reached a stable figure of around 5% for all patients who underwent a major resection for CRC during the last 4 years of the study period (Fig. 1). The Charlson comorbidity index of the overall study population increased over time with the proportion of patients with ≥1 comorbidity increasing from 22.8% to 25.2% during the course of this study period. FIGURE 1 Percentage of patients who receive a major resection for colorectal cancer, and who go on to have a hepatic resection, by the year of the primary colorectal resection. The data suggest that the proportion of liver resections carried out that were classified as major resections (hemihepatectomies and extended hemihepatectomies), reduced over the study period from 48.2% to 39.9% (P < 0.001, Supplementary Table 1).

FIGURE 1 Percentage of patients who receive a major resection for colorectal cancer, and who go on to have a hepatic resection, by the year of the primary colorectal resection. The data suggest that the proportion of liver resections carried out that were classified as major resections (hemihepatectomies and extended hemihepatectomies), reduced over the study period from 48.2% to 39.9% (P < 0.001, Supplementary Table 1). Variation in Resection Rates Across Providers The bar chart and organizational funnel plot (Fig. 2) show the variation in the rate of hepatic resection between different hospitals for the more recent time period, 2009 to 2012. There was a large degree of variation in the proportion of patients having a liver resection by the hospital where their primary tumor was treated, suggesting a difference in referral pathways between hospitals, with crude rates between 1.9% and 16.7% (P ≤ 0.001, Fig. 2A) and risk-adjusted rates ranging from 2.1% to 12.2% (Fig. 2B). Four hospitals were outliers (falling outside the 99.8% control limits on the funnel plot) with 2 having higher than expected rates of liver resection, and 2 having lower than expected rates. By Cancer Alliance (Supplementary Fig. 1), the crude rate ranged from 4.2% to 7.1% (P ≤ 0.001), with a similar risk-adjusted rate of 4.1% to –7.0%. FIGURE 2 Variation in the proportion of patients receiving a liver resection within three years, by the hospital carrying out their primary colorectal resection, (A) crude rate, (B) risk-adjusted funnel plot.

tant to note, however, that in this >70s group, liver resection rates have more than doubled over time, from 1.2% in 1998 to 2004 to 2.8% in 2005 to 2012 suggesting management patterns may have changed. The rate of major liver resection decreased over the period, from 48.2% to 33.8% of all liver resections carried out. Increasing deprivation was also associated with a reduced likelihood of receiving a liver resection11,20, with those in the most deprived quintile being 24% less likely to receive a resection than those in the most affluent quintile, as was increasing comorbidity.11,18 In addition, these data demonstrate that anatomical location of the CRC may influence the likelihood of liver resection as the odds of receiving a resection was lowest for right-sided primary tumors. Such an observation was also noted in 2 Swedish studies.2,18 There has also been a change in the management of patients with synchronous metastases (Stage IV at presentation), with the proportion of these who receive liver resection surgery >doubling from 8.0% for the period 1998 to 2004 to 18.9% in 2005 to 2012.11

Variation in Resection Rates Across Providers The bar chart and organizational funnel plot (Fig. 2) show the variation in the rate of hepatic resection between different hospitals for the more recent time period, 2009 to 2012. There was a large degree of variation in the proportion of patients having a liver resection by the hospital where their primary tumor was treated, suggesting a difference in referral pathways between hospitals, with crude rates between 1.9% and 16.7% (P ≤ 0.001, Fig. 2A) and risk-adjusted rates ranging from 2.1% to 12.2% (Fig. 2B). Four hospitals were outliers (falling outside the 99.8% control limits on the funnel plot) with 2 having higher than expected rates of liver resection, and 2 having lower than expected rates. By Cancer Alliance (Supplementary Fig. 1), the crude rate ranged from 4.2% to 7.1% (P ≤ 0.001), with a similar risk-adjusted rate of 4.1% to –7.0%. FIGURE 2 Variation in the proportion of patients receiving a liver resection within three years, by the hospital carrying out their primary colorectal resection, (A) crude rate, (B) risk-adjusted funnel plot. Analysis by the regional Cancer Alliance, within which the patient is treated for their primary colorectal resection shows that the rate of major liver resection (hemihepatectomy and extended hemihepatectomy) varied between 28% and 58% of all liver resections carried out (P < 0.001). This suggests there are regional differences in the surgical treatment of liver metastases.

he patient is treated for their primary colorectal resection shows that the rate of major liver resection (hemihepatectomy and extended hemihepatectomy) varied between 28% and 58% of all liver resections carried out (P < 0.001). This suggests there are regional differences in the surgical treatment of liver metastases. There are also differences in the type of liver resection carried out, depending on whether patients had their primary colorectal tumor removed at a hospital with a specialist liver center onsite. The major liver resection rate was 36.1% for liver centers compared to 41.1% for those treated for their colorectal primary in a hospital without a liver centre (P < 0.001) Variation in Resection Rates by Patient Characteristics Table 3 shows the results of the logistic regression model used to determine the odds of liver resection. There was no significant difference in the rate by year of primary resection in the adjusted model. The likelihood of liver resection decreased significantly with increasing age [odds ratio (OR) per 10-year increase 0.64; 95% confidence interval (CI), 0.62–0.65], with increasing deprivation level (OR for most deprived quintile compared to most affluent 0.76; 95% CI, 0.70–0.83)) and with increasing Charlson co-morbidity score (OR for Charlson score ≥3 compared to score of zero 0.43; 95% CI, 0.34–0.53). Women were less likely to receive a resection (OR 0.77; 95% CI, 0.73–0.81).

, with increasing deprivation level (OR for most deprived quintile compared to most affluent 0.76; 95% CI, 0.70–0.83)) and with increasing Charlson co-morbidity score (OR for Charlson score ≥3 compared to score of zero 0.43; 95% CI, 0.34–0.53). Women were less likely to receive a resection (OR 0.77; 95% CI, 0.73–0.81). The odds of resection were higher for tumors in the rectosigmoid (OR 1.90; 95% CI, 1.73–2.08), rectum (OR 1.60; 95% CI, 1.49–1.71) and left colon (OR 1.66; 95% CI, 1.56–1.77) compared to the right colon. The odds also increased with advancing stage of the disease at presentation (OR Stage IV vs Stage I 20.14; 95% CI, 17.60–23.04). In addition to the patient characteristics listed, elements of the treatment pathway affected the likelihood of resection. Patients who have their primary colorectal resection in a hospital with a liver center were more likely to receive a liver resection (OR 1.22; 95% CI; 1.10–1.35). This may be because of a higher proportion of liver resections taking place synchronously to their bowel resection for these patients (18%) compared to those receiving their bowel resection in a hospital without a specialist onsite liver center (3%). When the logistic regression model was run to only include patients at Stages I–III at diagnosis, this effect was observed in the unadjusted model (Supplementary Table 3, OR 1.10; 95% CI ,1.02–1.19) but was no longer significant following risk-adjustment, and was only seen for those patients with synchronous metastases (Supplementary Table 2).

ion model was run to only include patients at Stages I–III at diagnosis, this effect was observed in the unadjusted model (Supplementary Table 3, OR 1.10; 95% CI ,1.02–1.19) but was no longer significant following risk-adjustment, and was only seen for those patients with synchronous metastases (Supplementary Table 2). DISCUSSION This study presents the most up-to-date and largest reported population-based analysis of liver resection rates in CRC patients since Morris et al.11 It includes resection for both synchronous and metachronous metastatic disease and covers all patients receiving a major primary resection for CRC within the English NHS.

This study presents the most up-to-date and largest reported population-based analysis of liver resection rates in CRC patients since Morris et al.11 It includes resection for both synchronous and metachronous metastatic disease and covers all patients receiving a major primary resection for CRC within the English NHS. Despite a plateau being observed in the percentage of CRC patients undergoing a liver resection, significant differences in the likelihood of liver resection continue to be observed for different patient characteristics, hospitals, and Cancer Alliances. Women were less likely to receive a liver resection than men and this was independent of tumor location, stage, age, comorbidity, and deprivation quintile. Although the models used are not directly comparable, this reflects the findings of the previous analysis11 showing this trend has changed little over time. It has also been confirmed in several other studies.11,13,18 The lower likelihood of liver resection for women was present for all disease stages except for presentation at Stage I and so potential explanations such as differing surveillance or referral patterns between men and women should be explored in more detail. Older patients were still less likely to receive a liver resection,11,18–20 despite data showing they can achieve good outcomes.21 Although patients aged >70 years made up 51.4% of those undergoing major colorectal resection, they accounted for only 29.4% of those undergoing liver resection. It is important to note, however, that in this >70s group, liver resection rates have more than doubled over time, from 1.2% in 1998 to 2004 to 2.8% in 2005 to 2012 suggesting management patterns may have changed. The rate of major liver resection decreased over the period, from 48.2% to 33.8% of all liver resections carried out.

rs. Such an observation was also noted in 2 Swedish studies.2,18 There has also been a change in the management of patients with synchronous metastases (Stage IV at presentation), with the proportion of these who receive liver resection surgery >doubling from 8.0% for the period 1998 to 2004 to 18.9% in 2005 to 2012.11 Patients who received their major primary colorectal resection in a hospital designated in this study as a liver center were 22% more likely to receive a liver resection compared to those who had to be referred on elsewhere, suggesting that referral pathways may not be working as efficiently as they should be and patients are missing out on access to treatment depending on their proximity to a hospital with a liver center. This replicates findings from a study using HES data, which examined 4547 patients who underwent major bowel resection and had synchronous metastases at diagnosis; patients treated in a hospital with an onsite hepatobiliary center were more likely to receive a liver resection.13 Our study confirmed this in a population of people with both synchronous and metachronous metastases and found that the increased likelihood of liver resection for patients receiving their bowel resection in a hospital with an onsite specialist liver team was present for patients at Stages I–III at diagnosis in the nonadjusted model, but this was no longer significant following risk-adjustment. However patients with synchronous metastases were 43% more likely to receive a liver resection if they had their primary tumor resected in a hospital with an onsite liver center. Numerous studies have now shown that both surgeon and hospital specialization may be associated with improved surgical outcomes, especially for complex procedures.22–26 This study is an important contribution to that evidence base as, across the whole population, it suggests that proximity to a liver center is as important a factor as socioeconomic deprivation when it comes to the likelihood of receiving a liver resection.

ved surgical outcomes, especially for complex procedures.22–26 This study is an important contribution to that evidence base as, across the whole population, it suggests that proximity to a liver center is as important a factor as socioeconomic deprivation when it comes to the likelihood of receiving a liver resection. Around 35% of patients with Stage IV disease treated in a liver center received their liver resection before or at the same time as their major bowel resection, compared to 11% of patients in nonliver centers, showing a large degree of variation in the treatment pathway for synchronous metastases. Streamlining the referral process may lead to more patients being eligible for synchronous resection, which may improve outcomes. Although linkage of the cancer registry and HES data allows full coverage of all CRC patients, there are some limitations of the study. The routine nature of the data means that there is no information as to extent of liver disease, chemotherapy status, treatment intent, or patient choice so we do not know which patients were potentially eligible for liver resection but did not undergo surgery. However, it is unlikely that the burden of liver disease varies substantially across hospitals, once adjustment for other patient and tumor characteristics is performed and so the variation highlighted remains valid.

e do not know which patients were potentially eligible for liver resection but did not undergo surgery. However, it is unlikely that the burden of liver disease varies substantially across hospitals, once adjustment for other patient and tumor characteristics is performed and so the variation highlighted remains valid. In conclusion, a large degree of variation is still present in the rates of resection by hospital, even allowing for differences in age, stage, deprivation quintile, Charlson, tumor site, comorbidity index, year of primary resection, and whether the hospital has a liver center. This suggests that there is still variation in access to treatment for patients with both synchronous and metachronous hepatic disease and, in light of the poorer outcomes reported within the UK, such variation deserves further urgent investigation. Despite a plateau in liver resection rates, clinicians should not become complacent in the management of CRCLM patients. Significant variations have been observed in patient and hospital characteristics and the likelihood of patients receiving a liver resection. Likely there are multiple factors influencing this; however, further work is needed to elucidate the reasons for this variation, as addressing this has a significant potential to improve patient outcomes. Supplementary Material Supplemental Digital Content Sources of funding: This work is supported by the Bobby Moore Fund/Cancer Research UK (grant C23434/A23706) and Yorkshire Cancer Research (grant L394).

Despite a plateau in liver resection rates, clinicians should not become complacent in the management of CRCLM patients. Significant variations have been observed in patient and hospital characteristics and the likelihood of patients receiving a liver resection. Likely there are multiple factors influencing this; however, further work is needed to elucidate the reasons for this variation, as addressing this has a significant potential to improve patient outcomes. Supplementary Material Supplemental Digital Content Sources of funding: This work is supported by the Bobby Moore Fund/Cancer Research UK (grant C23434/A23706) and Yorkshire Cancer Research (grant L394). This work involves patient-level information collected by the NHS that has either been provided by, or derived from, patients as part of their care and support. The data are collated, maintained, and quality-assured by the National Cancer Registration and Analysis Service, which is part of Public Health England (PHE). Access to the data was facilitated by the PHE Office for Data Release. The authors report no conflicts of interests. DISCUSSANTS Marek Krawczyk (Warsaw, Poland): First of all, I would like to thank the European Surgical Association for giving me the opportunity to be the first discussant of such an interesting study. I have to congratulate the authors on the preparation of this study. They observed significant variation in the likelihood of liver resection, with liver resection being more likely for patients who had their primary tumors resected in a hospital with a specialist hepatobiliary service.

DISCUSSANTS Marek Krawczyk (Warsaw, Poland): First of all, I would like to thank the European Surgical Association for giving me the opportunity to be the first discussant of such an interesting study. I have to congratulate the authors on the preparation of this study. They observed significant variation in the likelihood of liver resection, with liver resection being more likely for patients who had their primary tumors resected in a hospital with a specialist hepatobiliary service. However, a few questions arise during the evaluation of the study: First, what was the effect in patients with complicated colon cancer (ie, perforation or bleeding) and liver metastases resected in the departments with or without liver surgery? Second, what was the influence of the number and size of liver metastases in the departments with or without liver surgery on the survival of the patients? Third, I’d like to know whether the chemical treatment has affected the performance of the resection of the liver with metachronic metastases. Perhaps, patients from wards that do not have liver surgery were only referred to chemotherapy without taking into account liver resection. I would be pleased to hear your answers to my questions.

Third, I’d like to know whether the chemical treatment has affected the performance of the resection of the liver with metachronic metastases. Perhaps, patients from wards that do not have liver surgery were only referred to chemotherapy without taking into account liver resection. I would be pleased to hear your answers to my questions. Response From Peter A. Lodge (Leeds, United Kingdom): Thank you for some really good questions. Obviously, one of the advantages of these datasets is the extensive population coverage. Unfortunately, with these administrative datasets, we do lose that clinical detail. We don’t know of any perforation or bleeding, and the effect that they may have. I can’t think of a population-based dataset in England, which would have that information. We are only able to identify the patients having colorectal surgery, but we’re not currently able to identify what complications they had before they had it. I think the same goes for the number, the size of the tumors, and the chemical treatment. These are clearly very important points to look at, but they are currently not recorded in the national dataset.

s having colorectal surgery, but we’re not currently able to identify what complications they had before they had it. I think the same goes for the number, the size of the tumors, and the chemical treatment. These are clearly very important points to look at, but they are currently not recorded in the national dataset. Massimo Malagò (London, United Kingdom): I have the privilege to work in the NHS and live in the United Kingdom, and I know and live the problems you mentioned in this interesting paper. Basically, it is very important that if there is an HPB-MDT or work in an environment close to an MDT, every case must be discussed at the regional HPB-MDT. Now, there is a article by Graham Poston in BJS from 2010, which also states that there are regional differences and different outcomes for HPB cancers throughout the UK. We tried to enforce the presentation of all patients with HPB cancers to specialist MDTs, but unfortunately, it seems that this does not work yet. So, what should we do to provide true equality of care in and beyond the UK?

es that there are regional differences and different outcomes for HPB cancers throughout the UK. We tried to enforce the presentation of all patients with HPB cancers to specialist MDTs, but unfortunately, it seems that this does not work yet. So, what should we do to provide true equality of care in and beyond the UK? Response From Peter A. Lodge (Leeds, United Kingdom): You make an important point. I think that what we’re trying to get at is that there are too many nonliver surgeons making decisions about what's operable or inoperable. Although there is an obligation for reporting the registration of primary hepatobiliary cancer in the UK, there's currently no obligation for the registration and MDT discussion of secondary CRC. Currently, I think that all we can do is go to the colorectal MDTs and encourage those physicians to send them in, unless we have very good links. However, there's always resistance from some of those groups. It's a difficult problem. I suspect that if we looked at something like recurrent CRC, we’d probably see a similar outcome. In Leeds, we have a high rate of dealing with recurrent CRC, but not all of the hospitals around will refer patients. It's an important topic, and trying to get the word out is key.

those groups. It's a difficult problem. I suspect that if we looked at something like recurrent CRC, we’d probably see a similar outcome. In Leeds, we have a high rate of dealing with recurrent CRC, but not all of the hospitals around will refer patients. It's an important topic, and trying to get the word out is key. Felice Giuliante (Rome, Italy): Thank you very much for your extremely interesting data and analysis. First, the issue lies in the lack of availability of liver surgeons in hospitals treating patients with CRCs; second, colorectal surgeons and medical oncologists should be made aware that the evaluation of the resectability of liver metastases should be done by dedicated liver surgeons, as this can change the clinical history of these patients. Of course, this is not only a problem for UK patients, but the fact that you clearly highlighted this important issue makes your study especially interesting. Do you think that a possible solution could be to create official networks, in which there are hospitals with colorectal surgery units and medical oncology units, which would then be connected to a specialized liver surgery unit in a referral hospital? This could be achieved through regular web conferences between different hospitals, during which official cases and imaging would be presented. In the UK, where liver surgery has been strongly centralized, this could be a way to face up to this problem of inequality in care. Could you comment on this?

eferral hospital? This could be achieved through regular web conferences between different hospitals, during which official cases and imaging would be presented. In the UK, where liver surgery has been strongly centralized, this could be a way to face up to this problem of inequality in care. Could you comment on this? Response From Peter A. Lodge (Leeds, United Kingdom): I think that's really important. Another piece of work we’ve done recently is to analyze colorectal MDTs within our region, to see whether we have access to all of their scans and so on, and verify what they should have referred. There were about one-third of patients, who they didn’t refer and we could have operated on. For this reason, resection rates remained low. It's about education and trying to get the word out.

Ts within our region, to see whether we have access to all of their scans and so on, and verify what they should have referred. There were about one-third of patients, who they didn’t refer and we could have operated on. For this reason, resection rates remained low. It's about education and trying to get the word out. Pierre-Alain Clavien (Zurich, Switzerland): Congratulations, Professor Lodge, for this important article. England is a role model for the centralization of complex procedures, with only a few centers performing major liver surgery. There was a similar article presented at the recent American Surgical Association meeting in Dallas, which reported on geographic variation in the utilization of liver resection in California with an effect on overall survival rate in CRC. This article was published in the September 2019 issue of Annals of Surgery. The issue in England seems to relate to the availability of MDTs, which correlates with the use of liver resection for colorectal liver metastases. How should this be fixed? With the mission of centralization, should large expert centers take the responsibility to organize MDTs with their own specialists, rather than just work in a network, expecting proper referrals? Should the government take the lead for such MDTs and better support the large centers for such a mission? I would love to have your input on this.

ralization, should large expert centers take the responsibility to organize MDTs with their own specialists, rather than just work in a network, expecting proper referrals? Should the government take the lead for such MDTs and better support the large centers for such a mission? I would love to have your input on this. Response From Peter A. Lodge (Leeds, United Kingdom): I think that this is a good question, but a difficult one to answer. The trouble is that there are local and national political influences. We can do our best to reach out to the MDT groups, which are not referring patients, but we have limited time and personnel. It certainly requires more investment. TABLE 1 OPCS 4.8 Codes for Major Colorectal Resection and Liver Resection

Response From Peter A. Lodge (Leeds, United Kingdom): I think that this is a good question, but a difficult one to answer. The trouble is that there are local and national political influences. We can do our best to reach out to the MDT groups, which are not referring patients, but we have limited time and personnel. It certainly requires more investment. TABLE 1 OPCS 4.8 Codes for Major Colorectal Resection and Liver Resection OPCS Code Description Primary colorectal resection H04 Total excision of colon and rectum H05 Total excision of colon H06 Extended excision of right hemicolon H07 Other excision of right hemicolon H08 Excision of transverse colon H09 Excision of left hemicolon H10 Excision of sigmoid colon H11 Other excision of colon H29 Subtotal excision of colon H33 Excision of rectum X14 Clearance of pelvis Liver resection J021 Right hemihepatectomy NEC J022 Left hemihepatectomy NEC J023 Resection of segment of liver J024 Wedge excision of liver J026 Extended right hemihepatectomy J027 Extended left hemihepatectomy J028 Other specified partial excision of liver J029 Partial excision of liver unspecified J031 Excision of lesion of liver J035 Excision of multiple lesions of liver J038 Other specified extirpation of lesion of liver J039 Unspecified extirpation of lesion of liver TABLE 2 Characteristics of the Study Population

ectomy J028 Other specified partial excision of liver J029 Partial excision of liver unspecified J031 Excision of lesion of liver J035 Excision of multiple lesions of liver J038 Other specified extirpation of lesion of liver J039 Unspecified extirpation of lesion of liver TABLE 2 Characteristics of the Study Population No Resection (n = 149,960) Liver Resection (n = 7423) Total (n = 157,383) Tumor site Colon 101,133 (67.4) 4584 (61.8) 105,717 (67.2) Cecum 24,307 (16.2) 846 (11.4) 25,153 (16) Appendix 1263 (0.8) 42 (0.6) 1305 (0.8) Ascending colon 14,585 (9.7) 481 (6.5) 15,066 (9.6) Hepatic flexure 4724 (3.2) 208 (2.8) 4932 (3.1) Transverse colon 8709 (5.8) 306 (4.1) 9015 (5.7) Splenic flexure 3640 (2.4) 189 (2.5) 3829 (2.4) Descending colon 4623 (3.1) 228 (3.1) 4851 (3.1) Sigmoid colon 33,601 (22.4) 2078 (28) 35,679 (22.7) Overlapping/unspecified lesion of colon 5681 (3.8) 206 (2.8) 5887 (3.7) Rectosigmoid 10,786 (7.2) 730 (9.8) 11,516 (7.3) Rectum 38,041 (25.4) 2109 (28.4) 40,150 (25.5) Sex Male 83,603 (55.8) 4660 (62.8) 88,263 (56.1) Female 66,357 (44.2) 2763 (37.2) 69,120 (43.9) Age at primary colorectal resection ≤60 28,511 (19) 2466 (33.2) 30,977 (19.7) 61–70 42,715 (28.5) 2777 (37.4) 45,492 (28.9) 71–80 51,157 (34.1) 1950 (26.3) 53,107 (33.7) >80 27,577 (18.4) 230 (3.1) 27,807 (17.7) Year of primary colorectal resection 2005 15,773 (10.5) 682 (9.2) 16,455 (10.5) 2006 17,676 (11.8) 805 (10.8) 18,481 (11.7) 2007 18,237 (12.2) 818 (11) 19,055 (12.1) 2008 19,116 (12.7) 891 (12) 20,007 (12.7) 2009 19,211 (12.8) 1016 (13.7) 20,227 (12.9) 2010 19,842 (13.2) 1069 (14.4) 20,911 (13.3) 2011 19,951 (13.3) 1084 (14.6) 21,035 (13.4) 2012 20,154 (13.4) 1058 (14.3) 21,212 (13.5) Tumor stage at diagnosis I 19,279 (12.9) 252 (3.4) 19,531 (12.4) II 49,202 (32.8) 1397 (18.8) 50,599 (32.2) III 47,330 (31.6) 2509 (33.8) 49,839 (31.7) IV 9587 (6.4) 2236 (30.1) 11,823 (7.5) Unknown 24,562 (16.4) 1029 (13.9) 25,591 (16.3) IMD quintile 1—least deprived 33,480 (22.3) 1820 (24.5) 35,300 (22.4) 2 34,022 (22.7) 1755 (23.6) 35,777 (22.7) 3 31,763 (21.2) 1528 (20.6) 33,291 (21.2) 4 27,886 (18.6) 1308 (17.6) 29,194 (18.5) 5—most deprived 22,809 (15.2) 1012 (13.6) 23,821 (15.1) Charlson Comorbidity Index 0 111,599 (74.4) 6099 (82.2) 117,698 (74.8) 1 25,917 (17.3) 1031 (13.9) 26,948 (17.1) 2 7668 (5.1) 215 (2.9) 7883 (5.0) ≥3 4776 (3.2) 78 (1.1) 4854 (3.1) Values in parentheses are percentages.

2) 4 27,886 (18.6) 1308 (17.6) 29,194 (18.5) 5—most deprived 22,809 (15.2) 1012 (13.6) 23,821 (15.1) Charlson Comorbidity Index 0 111,599 (74.4) 6099 (82.2) 117,698 (74.8) 1 25,917 (17.3) 1031 (13.9) 26,948 (17.1) 2 7668 (5.1) 215 (2.9) 7883 (5.0) ≥3 4776 (3.2) 78 (1.1) 4854 (3.1) Values in parentheses are percentages. TABLE 3 Odds of Having a Liver Resection Within 3 Years of Primary Colorectal Tumor Resection

2) 4 27,886 (18.6) 1308 (17.6) 29,194 (18.5) 5—most deprived 22,809 (15.2) 1012 (13.6) 23,821 (15.1) Charlson Comorbidity Index 0 111,599 (74.4) 6099 (82.2) 117,698 (74.8) 1 25,917 (17.3) 1031 (13.9) 26,948 (17.1) 2 7668 (5.1) 215 (2.9) 7883 (5.0) ≥3 4776 (3.2) 78 (1.1) 4854 (3.1) Values in parentheses are percentages. TABLE 3 Odds of Having a Liver Resection Within 3 Years of Primary Colorectal Tumor Resection Unadjusted Odds Ratio 95% Confidence Interval P P Across Groups Adjusted Odds Ratio 95% Confidence Interval P P Across Groups Year of resection of colorectal primary 1.03 1.02 1.05 <0.001 <0.001 1.00 1.00 1.02 0.155 0.155 Primary resection carried out in hospital with liver center 1.29 1.23 1.37 <0.001 <0.001 1.22 1.10 1.35 <0.001 <0.001 Age at resection of colorectal primary (per 10-y increase) 0.58 0.56 0.59 <0.001 <0.001 0.63 0.62 0.65 <0.001 <0.001 Sex <0.001 <0.001 Male 1.00 1.00 Female 0.75 0.71 0.78 <0.001 0.77 0.73 0.81 <0.001 IMD quintile <0.001 <0.001 1—least deprived 1.00 1.00 2 0.95 0.89 1.01 0.127 0.97 0.90 1.03 0.307 3 0.88 0.83 0.95 0.001 0.90 0.83 0.97 0.004 4 0.86 0.80 0.93 <0.001 0.87 0.81 0.94 <0.001 5—most deprived 0.82 0.75 0.88 <0.001 0.76 0.70 0.83 <0.001 Stage of primary tumor at diagnosis <0.001 <0.001 I 1.00 1.00 II 2.17 1.90 2.49 <0.001 2.66 2.32 3.04 <0.001 III 4.06 3.56 4.62 <0.001 4.46 3.91 5.09 <0.001 IV 17.84 15.63 20.37 <0.001 20.14 17.60 23.04 <0.001 unknown 3.21 2.79 3.68 <0.001 2.97 2.57 3.43 <0.001 Tumour site <0.001 <0.001 Right colon 1.00 1.00 Left colon 1.70 1.60 1.80 <0.001 1.66 1.56 1.77 <0.001 Rectosigmoid 1.92 1.76 2.10 <0.001 1.90 1.73 2.08 <0.001 Rectum 1.58 1.48 1.68 <0.001 1.60 1.49 1.71 <0.001 Colon unknown 1.03 0.89 1.19 0.674 1.06 0.91 1.23 0.469 Charlson comorbidity score <0.001 <0.001 0 1.00 1.00 1 0.73 0.68 0.78 <0.001 0.85 0.79 0.91 <0.001 2 0.51 0.45 0.59 <0.001 0.68 0.59 0.79 <0.001 ≥3 0.30 0.24 0.37 <0.001 0.43 0.34 0.53 <0.001

Bariatric surgery has been recognized as an effective and safe treatment option for morbid obesity and its related comorbidities.1–3 For the last 2 decades, Roux-en-Y gastric bypass has been considered as the treatment of choice for obesity World Health Organization grade III, with an acceptable complication rate, long-lasting weight control, and efficient reduction of comorbidities, especially type 2 diabetes (T2DM).4–6 The dramatically increasing prevalence of obesity has led to the development of alternative treatment strategies including new operative techniques, such as the laparoscopic sleeve gastrectomy (LSG), which was first described in 2003 by Regan et al7 in a staged concept. LSG is suggested to be a technically less demanding operation than laparoscopic Roux-en-Y gastric bypass (LRYGB), which offers some potential benefits over the LRYGB. Because the intestinal passage is still intact after LSG endoscopy of the remaining stomach and access to the duodenum is still possible, the risk of internal hernias is absent. In case of insufficient weight loss, LRYGB or biliopancreatic diversion with duodenal switch can be performed as a second-stage procedure. However, there has been some concern regarding onset or worsening of reflux and leakages after LSG, with the latter (although rare) being rather demanding to treat. LSG has gained popularity throughout the world, but there is still a lot of skepticism among bariatric surgeons regarding achievable long-term results in comparison with the established results of LRYGB. As a consequence, many studies in the past few years have been published assessing the midterm efficacy and safety of LSG.

s gained popularity throughout the world, but there is still a lot of skepticism among bariatric surgeons regarding achievable long-term results in comparison with the established results of LRYGB. As a consequence, many studies in the past few years have been published assessing the midterm efficacy and safety of LSG. The excessive weight loss ranged from 53% to 69%, and the incidence of staple line leaks and bleeding ranged from 1% to 3%.8–12 In 2012, the American Society for Metabolic and Bariatric Surgery published a revised position statement, which proposed that LSG is a valid alternative operation technique to LRYGB.13 However, currently, the evidence for the superiority of either surgical technique is still weak.14–16 Only 3 (2 from the same institution) randomized clinical trials comparing LSG and LRYGB with small patient numbers (16–30 per group) and limited follow-up (12–35 months) have been published so far. Although bariatric operations are currently among the most frequently performed operative procedures in the United States and Europe,17 a potential outcome difference between LSG and LRYGB would therefore have a substantial impact on the health care systems worldwide. Therefore, we decided to conduct a large multicentric randomized clinical trial assessing the efficacy and safety of LSG and LRYGB to answer this important question.

States and Europe,17 a potential outcome difference between LSG and LRYGB would therefore have a substantial impact on the health care systems worldwide. Therefore, we decided to conduct a large multicentric randomized clinical trial assessing the efficacy and safety of LSG and LRYGB to answer this important question. METHODS From January 2007 to November 2011, we enrolled 217 patients at 4 bariatric centers in Switzerland. The study protocol was reviewed and approved by the local ethical committees of each participating bariatric center and registered at the clinical trials registry of the National Institutes of Health (NCT 00356213). All patients were informed in detail about the potential risks and benefits of both operations and provided written informed consent for the participation of the study. Additional consent was obtained for the surgical procedure once the patient was randomized. The patients were evaluated by a multidisciplinary team (including a endocrinologist/physician, a psychiatrist, a nutritionist, and a bariatric surgeon) and were included in the study if they fulfilled the criteria for bariatric surgery in Switzerland [body mass index (BMI) >40 or >35 kg/m2 with the presence of at least 1 comorbidity, age between 18 and 65 years, and failure of conservative treatment over 2 years].

psychiatrist, a nutritionist, and a bariatric surgeon) and were included in the study if they fulfilled the criteria for bariatric surgery in Switzerland [body mass index (BMI) >40 or >35 kg/m2 with the presence of at least 1 comorbidity, age between 18 and 65 years, and failure of conservative treatment over 2 years]. In addition to general contraindications for major abdominal surgery, exclusion criteria were severe symptomatic gastroesophageal reflux disease (GERD) despite medication, large hiatal hernia, expected dense adhesions at the level of the small bowel, the need for endoscopic follow-up of the duodenum, and patients with inflammatory bowel disease. The primary end point of the study was weight loss, which was defined by excessive BMI loss (EBMIL), over a period of 5 years. To detect a 10% difference, we calculated a study size of 200 patients to reach a 94% power. This is an interim analysis at 1 year. Secondary end points were the rate of perioperative morbidity and mortality, the remission rates of the associated comorbidities, and the change in quality of life (QOL) in the 2 patient groups.

ect a 10% difference, we calculated a study size of 200 patients to reach a 94% power. This is an interim analysis at 1 year. Secondary end points were the rate of perioperative morbidity and mortality, the remission rates of the associated comorbidities, and the change in quality of life (QOL) in the 2 patient groups. Randomization Patients were assigned to either the LSG or LRYGB group, using a computer-based randomization with sealed envelopes after informed consent has been obtained in the outpatient clinic. Using this strategy, a total of 225 patients were eligible and randomized. Eight patients were excluded because of the following reasons: 1 patient crossed over from the LRYGB to LSG group because of unexpected dense adhesions of the jejunum, and 7 patients were operated on after November 2011 when the recruitment phase was closed. This resulted in a total of 217 patients who were included in the study.

nts were excluded because of the following reasons: 1 patient crossed over from the LRYGB to LSG group because of unexpected dense adhesions of the jejunum, and 7 patients were operated on after November 2011 when the recruitment phase was closed. This resulted in a total of 217 patients who were included in the study. Operation Techniques All patients were operated on using standardized operation techniques. For LSG, we used a 35-Fr bougie along the lesser curvature for calibration of the gastric tube; the longitudinal resection of the stomach was done from approximately 3 to 6 cm orally of the pylorus to the angle of His. The staple line was routinely oversewn with an absorbable running suture. Hiatal hernias were explored and repaired with posterior closure of the crura. LRYGB was performed with a 150-cm antecolic Roux-limb with either a linear stapled or circular stapled (25-mm) gastrojejunostomy according to the preference of the bariatric center. A 50-cm long biliopancreatic limb was chosen.

nning suture. Hiatal hernias were explored and repaired with posterior closure of the crura. LRYGB was performed with a 150-cm antecolic Roux-limb with either a linear stapled or circular stapled (25-mm) gastrojejunostomy according to the preference of the bariatric center. A 50-cm long biliopancreatic limb was chosen. Assessment All patients underwent complete evaluation before the operation (including endoscopy, esophageal manometry, upper gastrointestinal series, and abdominal ultrasonography). Additional investigations were performed according to the risk profile of each individual patient. Synchronous cholecystectomy was performed according to the center policy for patients with gallstones. Patients were followed at each center according to a protocol, 4 times during the first postoperative year and yearly intervals thereafter. Eating behavior, comorbidities, anthropometric parameters, clinical parameters including blood samples, and QOL using the Gastrointestinal Quality of Life Index (GIQLI) were routinely assessed.18 Vitamin supplementation and postoperative thrombosis prophylaxis were performed according to the center policy. Comorbidities were defined using international standard criteria (hypertension: systolic blood pressure 140 mm Hg or more and/or diastolic blood pressure ≥90 mm Hg or antihypertensive drug therapy; diabetes mellitus type 2 (T2DM): fasting plasma glucose ≥126 mg/dL or 2-hour plasma glucose ≥200 mg/dL during oral glucose tolerance test or antidiabetic drug with or without insulin therapy; impaired glucose tolerance: 2-hour plasma glucose ≥140 and ≤200 mg/dL during oral glucose tolerance test; dyslipidemia: fasting high-density lipoprotein <40 mg/dL for men, <50 mg/dL for women, and/or triglycerides >150 mg/dL and/or low-density lipoprotein >100 mg/dL or the use of statins; obstructive sleep apnea syndrome: repeated upper airway occlusions during sleep with or without sleepiness and high apnea/hypopnea index and the need for continuous positive airway pressure during sleep; and GERD: need for proton pump inhibitor agents and/or esophagitis diagnosed on endoscopy and/or abnormal manometry; athralgia: clinical and radiological findings). Remission and improvement of comorbidities were defined by the endocrinologist/physician responsible for the follow-up. The definitions for remission of diabetes from the International Diabetes Association were not published yet at the time of the submission of the protocol.

ralgia: clinical and radiological findings). Remission and improvement of comorbidities were defined by the endocrinologist/physician responsible for the follow-up. The definitions for remission of diabetes from the International Diabetes Association were not published yet at the time of the submission of the protocol. Perioperative complications were defined using a standardized complication classification system, which has been shown to be very reliable.19 In brief, the Clavien-Dindo classification for grading the severity of complications by the therapy used to correct a specific complication. Grade I complications include minor deterioration from the normal postoperative course. Grade II complications can be treated by drugs, blood transfusion, physiotherapy, or nutritional support. Grade III complications require interventional (IIIa) or operative (IIIb) treatment. Grade IV complications are life-threatening complications with the need of intensive care unit management. Grade V is defined as death of the patient. If a patient had more than 1 complication, only the highest ranked complication was used for the analysis. Statistical Analysis Data analysis was performed using IBM SPSS for Windows (version 21; IBM, Armonk, NY). Values are reported as means ± SD. Descriptive statistics were used for demographic variables such as age, weight, and BMI. The Student t and Fisher exact 2-sided tests were used where appropriate. A P value of less than 0.05 was considered as statistically significant.

M SPSS for Windows (version 21; IBM, Armonk, NY). Values are reported as means ± SD. Descriptive statistics were used for demographic variables such as age, weight, and BMI. The Student t and Fisher exact 2-sided tests were used where appropriate. A P value of less than 0.05 was considered as statistically significant. RESULTS Patient Characteristics A total of 217 patients with a BMI between 35 and 61 kg/m2 were randomized to either LSG (107 patients) or LRYGB (110 patients). The follow-up rate was 100% at 1 year. One hundred twelve patients completed the 2-year follow-up and 70 patients the 3-year follow-up at the time of analysis (median follow-up of 2 years). The 2 groups were similar in terms of sex distribution, age, weight, BMI, and QOL (Table 1). There were no differences regarding eating disorder/behavior. The rate of comorbidities, such as diabetes, hypertension, hyperlipidemia, obstructive sleep apnea syndrome, and others, were almost identical in both groups (Table 1).

2 groups were similar in terms of sex distribution, age, weight, BMI, and QOL (Table 1). There were no differences regarding eating disorder/behavior. The rate of comorbidities, such as diabetes, hypertension, hyperlipidemia, obstructive sleep apnea syndrome, and others, were almost identical in both groups (Table 1). TABLE 1. Patient Descriptives Perioperative Data/Complications The mean operative time was less for LSG than for LRYGB (87.2 ± 52.3 minutes vs 108 ± 42.3; P = 0.003). The conversion rate was 0.9% in both groups. In the LSG group, 1 patient had to be converted to an open procedure because of technical difficulty with extreme intra-abdominal fat and enormous liver size. In the LRYGB group, 1 patient was converted to an open procedure to control bleeding at the gastric remnant. Additional operations have been performed in 36 of 107 patients in the LSG group and 26 of 110 patients in the LRYGB group (P = 0.09). Among these, the main operations were cholecystectomies (15 in the LSG group and 19 in the LRYGB group) and hiatal hernia repair (13 in the LSG group and 1 in the LRYGB group; P = 0.001).

ditional operations have been performed in 36 of 107 patients in the LSG group and 26 of 110 patients in the LRYGB group (P = 0.09). Among these, the main operations were cholecystectomies (15 in the LSG group and 19 in the LRYGB group) and hiatal hernia repair (13 in the LSG group and 1 in the LRYGB group; P = 0.001). The perioperative complication rate (<30 days) assessed by the Clavien-Dindo classification was higher in the LRYGB group than in the LSG group (17.2% vs 8.4%; P = 0.067). The grading of the complications is given in Table 2. The rate of severe complications requiring a reoperation was 4.5% (5/110) in the LRYGB group versus 0.9% (1/107) in the LSG group (P = 0.21). The reason for the reoperation in the LSG group was obstruction of the gastric sleeve. The reasons for the 5 revisions in the LRYGB group were as follows: 1 leakage at the gastrojejunostomy, 1 obstruction of the biliopancreatic limb, 2 intra-abdominal abscesses, and 1 pleural empyema. TABLE 2. Perioperative Morbidity The different complications stratified into minor and major complications are given in Table 2. One patient had an early leakage at the gastrojejunostomy after LRYGB, with a complicated course including aspiration, acute respiratory distress syndrome, multiorgan failure, and finally an episode of intracerebral bleeding, which eventually led to the death of the patient.

nd major complications are given in Table 2. One patient had an early leakage at the gastrojejunostomy after LRYGB, with a complicated course including aspiration, acute respiratory distress syndrome, multiorgan failure, and finally an episode of intracerebral bleeding, which eventually led to the death of the patient. Weight Loss at 1, 2, and 3 Years There was a significant weight loss at 1 year in both groups. In LSG patients, the weight loss was from 123.5 ± 19.4 to 86.9 ± 16.9 kg compared with the weight loss in LRYGB patients from 124.7 ± 19.8 to 84.7 ± 16.8 kg. There was no difference regarding weight loss or EBMIL between the 2 groups after 1 year (Figs. 1A, B), and there was no further weight loss in patients who completed the follow-up at 2 (n = 112) and 3 (n = 70) years. We observed a tendency toward a lower EBMIL at 3 years in the LSG group (n = 38) than in the LRYGB group (n = 32) [63.3% vs 72.8% (NS)]. FIGURE 1. A, Change in BMI (means ± standard error). B, EBMIL (means).

Weight Loss at 1, 2, and 3 Years There was a significant weight loss at 1 year in both groups. In LSG patients, the weight loss was from 123.5 ± 19.4 to 86.9 ± 16.9 kg compared with the weight loss in LRYGB patients from 124.7 ± 19.8 to 84.7 ± 16.8 kg. There was no difference regarding weight loss or EBMIL between the 2 groups after 1 year (Figs. 1A, B), and there was no further weight loss in patients who completed the follow-up at 2 (n = 112) and 3 (n = 70) years. We observed a tendency toward a lower EBMIL at 3 years in the LSG group (n = 38) than in the LRYGB group (n = 32) [63.3% vs 72.8% (NS)]. FIGURE 1. A, Change in BMI (means ± standard error). B, EBMIL (means). Resolution of Comorbidities The rate of comorbidities improved dramatically in both groups 1 year after the operation. Figure 2 displays the percentage of patients who were cured or showed improvement in their comorbidities. For patients with T2DM, 57.7% in the LSG group and 67.9% in the LRYGB group were not taking medication. Except for the remission of GERD, there was no difference between the LSG group and the LRYGB group regarding the remission of comorbidities or improvement rate. Patients undergoing LSG experienced a slightly higher rate of new-onset GERD (12.5% vs 4%; P = 0.12), and among those who already presented with GERD before the operation, the rate of improvement was significantly lower than those who underwent LRYGB (50% vs 75%; P = 0.008). The QOL assessed at 1 year was equal between patients undergoing LSG and those undergoing LRYGB, with 127 and 128 points, respectively, at 1 year (NS). Patients from both groups experienced a significant improvement in QOL compared with baseline (P < 0.0001) and even exceeded that of healthy individuals who reach a score of 121 points (P < 0.01).18

equal between patients undergoing LSG and those undergoing LRYGB, with 127 and 128 points, respectively, at 1 year (NS). Patients from both groups experienced a significant improvement in QOL compared with baseline (P < 0.0001) and even exceeded that of healthy individuals who reach a score of 121 points (P < 0.01).18 FIGURE 2. Reduction in comorbidity 1 year after surgery. No significant difference in cure or improvement of comorbidities between LSG and LRYGB except for GERD (*P = 0.008). GERD indicates gastro esophageal reflux disease; OSAS, obstructive sleep apnea syndrome; T2DM, type 2 diabetes. Complications During the First Postoperative Year In the LRYGB group, there was 1 anastomotic ulcer at the gastroenterostomy and 1 stricture that needed endoscopic dilatation. Up to 1 year postoperatively, no patient had to be reoperated on for either insufficient weight loss or internal hernia in both groups. Two patients of the LSG group experienced severe GERD symptoms, but until 1 year after the operation, none of them agreed to have undergone conversion to LRYGB. Both patients were asymptomatic and not receiving proton pump inhibitor therapy before the operation. The incidence of micronutrient deficiency was equal in both groups (LSG: n = 28 patients; LRYGB: n = 27 patients), with vitamin D deficiency being the most frequent deficiency, followed by vitamin B12 deficiency (LSG: n = 7; LRYGB: n = 15; P < 0.12). Iron, folic acid, and zinc were rarely deficient.

fore the operation. The incidence of micronutrient deficiency was equal in both groups (LSG: n = 28 patients; LRYGB: n = 27 patients), with vitamin D deficiency being the most frequent deficiency, followed by vitamin B12 deficiency (LSG: n = 7; LRYGB: n = 15; P < 0.12). Iron, folic acid, and zinc were rarely deficient. DISCUSSION This is the first multicentric randomized clinical trial including a large number of patients with 100% 1-year follow-up comparing LSG versus LRYGB. We found that LSG and LRYGB are almost equally efficient regarding weight loss and reduction in comorbidities. There was a trend toward fewer perioperative complications in patients undergoing LSG. However, the difference was not statistically significant.

patients with 100% 1-year follow-up comparing LSG versus LRYGB. We found that LSG and LRYGB are almost equally efficient regarding weight loss and reduction in comorbidities. There was a trend toward fewer perioperative complications in patients undergoing LSG. However, the difference was not statistically significant. LSG can be considered a technically less demanding operation than the LRYGB, which is reflected by the shorter operation time and the lower complication rate in the LSG group. Our results of the perioperative morbidity are comparable with those of a recently published Finish trial, which focused only on the perioperative complication rate between LSG and LRYGB.20 That study reported a reduced operating time and a lower overall complication rate in the LSG group. The complication rate was slightly higher, with 13.2% for LSG and 26.5% for Roux-en-Y gastric bypass than ours. In line with our results, there was no difference regarding major complications. Despite the lower complication rate, it needs to be emphasized that LSG is not a harmless operation, because it can be associated with deleterious complications. Leakage at the gastroesophageal junction or (less frequently) in the antrum is known to be very unpleasant and cumbersome to treat.21 The leakage rate is between 1% and 3%22 but can be as high as 16% in reoperations.23 Mild dysphagia after LSG can be observed frequently, but strictures or torsions of the gastric sleeve are complications that are difficult to treat and often result in the resection of the gastric sleeve at the end of the treatment line. Therefore, it is utmost important that this procedure is performed with the best standardized technique by experienced bariatric surgeons.

strictures or torsions of the gastric sleeve are complications that are difficult to treat and often result in the resection of the gastric sleeve at the end of the treatment line. Therefore, it is utmost important that this procedure is performed with the best standardized technique by experienced bariatric surgeons. In the past, there has been skepticism regarding LSG and GERD, because the anatomical structure of the angle of His is no longer intact after LSG. Furthermore, there is still a large proportion of remaining parietal cells. Accordingly, the new-onset rate of GERD has been reported as high as 21% after LSG.10 In line with this, we observed a significantly lower rate of GERD remission and a clear trend of new-onset GERD after LSG compared with LRYGB. Therefore, we believe that patients with preexisting GERD are at a risk of deterioration after LSG and should rather undergo LRYGB. This is one result of our study that allows the bariatric surgeon to tailor the operative strategy for the patient according to his or her preoperative risk profile. We also propose that patients undergo LSG if they have expected major adhesions, need a staged concept, or suffer from Crohn disease. It is hoped that such a tailored approach will optimize the long-term results of bariatric surgery.

operative strategy for the patient according to his or her preoperative risk profile. We also propose that patients undergo LSG if they have expected major adhesions, need a staged concept, or suffer from Crohn disease. It is hoped that such a tailored approach will optimize the long-term results of bariatric surgery. The strengths of this study are the standardized multicentric design and the complete follow-up of all included patients at 1 year, reflecting a representative patient population. Both groups were almost identical regarding the anthropometric values and the prevalence of comorbidities, and both procedures led to a similar and significant weight loss comparable with previous nonrandomized studies.14 At 3 years, EBMIL after LSG was slightly lower than that of the LRYGB group. However, only a third of all patients reached this time point so far.

anthropometric values and the prevalence of comorbidities, and both procedures led to a similar and significant weight loss comparable with previous nonrandomized studies.14 At 3 years, EBMIL after LSG was slightly lower than that of the LRYGB group. However, only a third of all patients reached this time point so far. One of the main goals of bariatric surgery is the long-term remission of the obesity-associated comorbidities such as T2DM, hypertension, and dyslipidemia. In our study, the rate of remission/improvement of the comorbidities was very high in both treatment groups, confirming the metabolic effect of both LSG and LRYGB. The potential different mechanisms behind the 2 surgical techniques have been investigated by 2 participating centers of this article in the past.16,24–27 Interestingly, we did not observe a difference between the 2 procedures regarding T2DM resolution. T2DM was efficiently treated by both procedures, which is important because T2DM is one of the main health care cost factors in obese patients.28 Further follow-up results of this study will have to be awaited for a firm statement regarding long-term remission rates of comorbidities and weight loss between the 2 procedures.

was efficiently treated by both procedures, which is important because T2DM is one of the main health care cost factors in obese patients.28 Further follow-up results of this study will have to be awaited for a firm statement regarding long-term remission rates of comorbidities and weight loss between the 2 procedures. CONCLUSIONS LSG can be performed faster than LRYGB and is associated with fewer perioperative complications. We could show that LSG and LRYGB are equally efficient regarding weight loss, reduction in comorbidities, and increase in QOL at 1 year. Therefore, we believe that LSG is a valuable surgical alternative for selected patients with morbid obesity. Long-term follow-up data are needed to confirm these results. ACKNOWLEDGMENTS The authors thank Phillip Hendrickson and Sandra Gagliardo for data management. Reprints: Ralph Peterli, MD, Department of Surgery, Claraspital, Kleinriehenstrasse 30, CH-4016 Basel, Switzerland. E-mail: ralph.peterli@claraspital.ch. Disclosure: SM-BOSS was supported by the Swiss National Science Foundation (SNF grant 32003B-120020) and Ethicon Endo Surgery USA. The authors declare no conflicts of interest.

Reprints: Ralph Peterli, MD, Department of Surgery, Claraspital, Kleinriehenstrasse 30, CH-4016 Basel, Switzerland. E-mail: ralph.peterli@claraspital.ch. Disclosure: SM-BOSS was supported by the Swiss National Science Foundation (SNF grant 32003B-120020) and Ethicon Endo Surgery USA. The authors declare no conflicts of interest. DISCUSSANTS M. Morino (Torino, Italy): I would like to thank the Society for giving me the opportunity to discuss this article and Dr Peterli for giving me the opportunity to review the manuscript. Dr Peterli and coworkers have to be complimented on this RCT; it addresses one of the most interesting topics on bariatric surgery at present, the comparison between sleeve and bypass. In fact your study was designed and powered on the basis of long-term (5-year) excess weight loss. You did not mention that in this presentation, but this I think is quite important. You have already published twice on different aspects of the early results of this trial, and today you presented the results at 1, 2, and 3 years, but it is important to note that less than 30% of the patients have reached the 3-year end point and less than 50% have reached the 2-year end point. Keeping in mind that the design of the study was powered at 5 years, my first question is whether you believe that the data that you presented today are sufficiently powered from a statistical point of view to sustain your conclusions?

DISCUSSANTS M. Morino (Torino, Italy): I would like to thank the Society for giving me the opportunity to discuss this article and Dr Peterli for giving me the opportunity to review the manuscript. Dr Peterli and coworkers have to be complimented on this RCT; it addresses one of the most interesting topics on bariatric surgery at present, the comparison between sleeve and bypass. In fact your study was designed and powered on the basis of long-term (5-year) excess weight loss. You did not mention that in this presentation, but this I think is quite important. You have already published twice on different aspects of the early results of this trial, and today you presented the results at 1, 2, and 3 years, but it is important to note that less than 30% of the patients have reached the 3-year end point and less than 50% have reached the 2-year end point. Keeping in mind that the design of the study was powered at 5 years, my first question is whether you believe that the data that you presented today are sufficiently powered from a statistical point of view to sustain your conclusions? The second point is that 200 patients randomized at 4 centers in 5 years means 10 patients per center per year. If you divide for 1 or 2 surgeons who probably perform this surgery in your group or maybe even 3, it makes 1 patient every 3 months. So, do you believe that this limited accrual is related to a limited rate of accrual or to a high exclusion rate?

andomized at 4 centers in 5 years means 10 patients per center per year. If you divide for 1 or 2 surgeons who probably perform this surgery in your group or maybe even 3, it makes 1 patient every 3 months. So, do you believe that this limited accrual is related to a limited rate of accrual or to a high exclusion rate? My third question concerns the GERD. In the manuscript, it is a little bit confusing, as you state that the symptoms of GERD were an exclusion criterion and then you state that in the sleeve group, the rate of GERD improvement was lower than that in the bypass group. Could you please comment on this point? Response From R. Peterli (Basel, Switzerland): It is true that the study was designed for an end point at 5 years, but as the procedures are gaining such popularity, we believe that it is important to address these results at 1 year and it is only the weight loss results that we have presented at 2 and 3 years. There are very few randomized studies in bariatric surgery, and it is a general problem in surgery that we do not have randomized trials. In the United States, you often have only 1-year results and it is very rare to have more than 25% follow-up rate at 5 years. If you have 25% follow-up rate at 5 years, one does not know the majority of the results. So we thought that it was important to address these results because this operation is gaining such popularity.

United States, you often have only 1-year results and it is very rare to have more than 25% follow-up rate at 5 years. If you have 25% follow-up rate at 5 years, one does not know the majority of the results. So we thought that it was important to address these results because this operation is gaining such popularity. It is true that it took us a long time to recruit. As the principal investigator, I have experience with duodenal switch, so I knew the technique. The sleeve gastrostomy technique had to be standardized and the other centers started later on. So there was a certain delay until we had it standardized for the other centers, and ethical approval in the other centers also took time. That explains the differences: the small number per year, per center the other centers that started later on. We did exclude severe GERD and big hiatal hernias, but many of these patients have some GERD symptoms; indeed, 48% of our patients did have intermittent reflux symptoms, so we did not exclude them, we excluded only severe GERD.

e-specific survival was not reached for AC and MAC. For SC, median cause-specific survival was 10 months (95% CI, 2–19 months). Compared with AC, cause-specific survival was significantly reduced for MAC (P = 0.011) and SC (P < 0.001). Furthermore, cause-specific survival was worse for SC compared with MAC (P < 0.001). Figure 1 displays the Kaplan-Meier survival curves for the different histological subtypes. Within the groups AC, MAC, and SC, there were no significant differences in survival between patients with colon cancer and rectal cancer, although patients with MAC of the rectum showed a trend toward shorter survival than patients with MAC of the colon (P = 0.065, Fig. 2). The cause-specific 5-year survival was 64 ± 3% for MAC of the colon (n = 270), 56 ± 8% for MAC of the rectum, which had received neoadjuvant treatment (n = 40), and 52 ± 7% for MAC of the rectum without neoadjuvant treatment (n = 63). Five-year survival did not differ significantly among these groups (P = 0.108). FIGURE 1. Cause-specific survival for patients of all stages, corresponding to adenocarcinoma, mucinous adenocarcinoma, and signet-ring cell carcinoma. FIGURE 2. Cause-specific survival for patients with (A) adenocarcinoma, (B) mucinous adenocarcinoma, and (C) signet-ring cell carcinoma, corresponding to colon cancer and rectal cancer.

It is true that it took us a long time to recruit. As the principal investigator, I have experience with duodenal switch, so I knew the technique. The sleeve gastrostomy technique had to be standardized and the other centers started later on. So there was a certain delay until we had it standardized for the other centers, and ethical approval in the other centers also took time. That explains the differences: the small number per year, per center the other centers that started later on. We did exclude severe GERD and big hiatal hernias, but many of these patients have some GERD symptoms; indeed, 48% of our patients did have intermittent reflux symptoms, so we did not exclude them, we excluded only severe GERD. DISCUSSANTS F. Pattou (Lille, France): Thank you very much for this discussion and for this important randomized study in a field in which there are relatively few. This operation in France has become, as in many countries, the most performed operation and that raises concern, at least in France. You talked to us about the safety; I will not come back to the long-term concern, which is a major one, but come back to the safety. You concluded that this operation is safer than gastric bypass. I would like to ask you about a complication that is now being seen with increasing frequency, acute leak at the top of the suture line. It may occur in only a few percent of patients, perhaps the reason you did not mention it, but it is extremely worrisome in everyday clinical life. In France, a large national study showed that these patients have at least 2 months' mean hospital stay. Could you please comment on this specific point of safety and the specific complication of acute leak?

of patients, perhaps the reason you did not mention it, but it is extremely worrisome in everyday clinical life. In France, a large national study showed that these patients have at least 2 months' mean hospital stay. Could you please comment on this specific point of safety and the specific complication of acute leak? Response From R. Peterli (Basel, Switzerland): Thank you, Prof Pattou, for this is a very important remark. It is true that the leak rate, in a consensus meeting last year, was 1% and good centers have 0% to 0.5%. However, that is not the issue, it is how do you perform the operation, what kind of staple is used. If you standardize the technique, you can decrease the leakage rate. But the problem is how do you manage a leak. Because it seems such an easy operation, many surgeons start doing it. However, they do not have experience in bariatric surgery and they start doing sleeves because they think it is easier to perform than a bypass and that is the danger. That is why I think it is important to give this message. I think it is true that if a leak occurs, it is a catastrophe and it has to be managed properly. The centers that start doing this type of surgery must work together with experienced people. They can start, but then they must have a contact, so if they have a leak, they know how to manage it. It is an issue, and I say that the sleeve is safer in terms of frequency of complications and that is how we powered the study also, as a secondary end point. But obviously, the leak, if it occurs, is much easier to treat in the bypass patients; that is true.

As early as 1979, the World Health Organization introduced the classification of colorectal cancers according to their histology. Histological subtypes were defined as classical adenocarcinomas (AC), which account for the large majority of cases, and mucinous adenocarcinomas (MAC), signet-ring cell carcinomas (SC), and other even less frequent forms (small cell carcinoma, squamous cell carcinoma, adenosquamous carcinoma, medullary carcinoma, and undifferentiated carcinoma).1,2 This classification exists in addition to the commonly used TNM staging and grading system and remains valid today.2 The histological subtype presumably plays a role in tumor biology and prognosis.3–6 However, in contrast to staging and grading of colorectal cancer, the histological subtype is currently not incorporated in clinical classification systems, although it potentially represents entities with different biological behavior, aggressiveness, and prognosis.4–7

y plays a role in tumor biology and prognosis.3–6 However, in contrast to staging and grading of colorectal cancer, the histological subtype is currently not incorporated in clinical classification systems, although it potentially represents entities with different biological behavior, aggressiveness, and prognosis.4–7 In general, about 10% of all colorectal cancers are MAC, and about 1% are SC.3,8–10 Because of their relatively rare occurrence, in particular, the evaluation of the clinical impact of SC is difficult. However, compared with AC, both MAC and SC have been shown to be associated with young age, advanced tumor stage, accumulation in female patients, and distinct molecular patterns, such as microsatellite instability and activating mutations of the BRAF gene.3,4,8,11 Although ambiguous, recent data and meta-analyses suggest that the histological subtype MAC may be associated with worse outcome compared with AC.3,4,9,12–14 Poor prognosis of SC is more evident, mainly due to high rates of synchronous and metachronous distant organ metastasis associated with this histological subtype.3,8,11 The purpose of this study was to characterize patients with colorectal MAC and SC through evaluation of a large institution-based cohort.

FIGURE 1. Cause-specific survival for patients of all stages, corresponding to adenocarcinoma, mucinous adenocarcinoma, and signet-ring cell carcinoma. FIGURE 2. Cause-specific survival for patients with (A) adenocarcinoma, (B) mucinous adenocarcinoma, and (C) signet-ring cell carcinoma, corresponding to colon cancer and rectal cancer. Cancer Subtypes and Prognosis In addition to the histological subtype of cancer (AC, MAC, or SC), factors associated with poor cause-specific survival on univariable analysis were multivisceral surgery (P < 0.001), tumor, node, metastasis and UICC stage (P < 0.001, respectively), poor differentiation (G3/4, P < 0.001), lymphatic invasion (P < 0.001), angioinvasion (P < 0.001), resection status other than R0 (P < 0.001), and recurrence of disease (P < 0.001). On multivariable analysis, most of these parameters remained independent prognostic factors, with the exception of grading, lymphatic invasion, and angioinvasion (Table 3). MAC and SC were associated significantly more often with higher tumor stages than AC. After correction for tumor stage, the difference in survival between MAC and AC was no longer significant (HR = 1.0, 95% CI = 0.83–1.2; P = 0.991; Table 3). However, SC remained an independent predictor of poor prognosis, with an HR of 2.5 (95% CI = 1.6–3.8; P < 0.001). The stage-corrected survival characteristics confirmed these findings (Table 4 and Fig. 3).

In general, about 10% of all colorectal cancers are MAC, and about 1% are SC.3,8–10 Because of their relatively rare occurrence, in particular, the evaluation of the clinical impact of SC is difficult. However, compared with AC, both MAC and SC have been shown to be associated with young age, advanced tumor stage, accumulation in female patients, and distinct molecular patterns, such as microsatellite instability and activating mutations of the BRAF gene.3,4,8,11 Although ambiguous, recent data and meta-analyses suggest that the histological subtype MAC may be associated with worse outcome compared with AC.3,4,9,12–14 Poor prognosis of SC is more evident, mainly due to high rates of synchronous and metachronous distant organ metastasis associated with this histological subtype.3,8,11 The purpose of this study was to characterize patients with colorectal MAC and SC through evaluation of a large institution-based cohort. PATIENTS AND METHODS Patients Since 1982, all patients undergoing surgery for colorectal cancer at the Department of Surgery, Klinikum rechts der Isar, Technische Universität München, Munich, Germany, are scheduled for periodic follow-up at our interdisciplinary ambulatory tumor center or outside of our hospital according to the recommendations of the German Cancer Society. The recommendations include—over a period of at least 5 years—regular physical examination, blood analysis, determination of carcinoembryonic antigen level, abdomen ultrasonography/computed tomography, chest radiography, and colonoscopy. All patient data are prospectively entered in a database, including preoperative tumor staging, preoperative multimodal treatment, details of the surgical procedure, occurrence of complications, postoperative histopathology, application of adjuvant or palliative treatment, and follow-up (date of last visit, date and site of tumor recurrence, date of tumor-related or unrelated death, cause-specific and recurrence-free survival).15 Information from patients followed outside of our institution is obtained by periodic phone calls to the responsible general practitioners or gastroenterologists.

d follow-up (date of last visit, date and site of tumor recurrence, date of tumor-related or unrelated death, cause-specific and recurrence-free survival).15 Information from patients followed outside of our institution is obtained by periodic phone calls to the responsible general practitioners or gastroenterologists. For this analysis, consecutive complete data sets of patients with resection of colorectal AC, MAC, or SC were extracted. Patients with other histological subtypes and patients without oncological resection of their primary tumor were not included. The latest date of inclusion and follow-up was October 1, 2012. Histological, clinical, and survival data of patients with classical AC were compared with those of patients with MAC and SC. All patients were staged according to the seventh edition of the International Union Against Cancer (UICC)/American Joint Committee on Cancer (AJCC) tumor staging system. Pathological Examination After tumor resection, each surgical specimen was fixed in 10% formalin, opened and thoroughly examined macroscopically. Multiple representative slices of 3 to 5 mm were obtained from the tumor, the junction with adjacent uninvolved tissue, proximal and distal resection margins, uninvolved colon, and macroscopically abnormalities if any, such as polyps or other morphological changes. The mesenteric fat was thoroughly examined for lymph nodes. The obtained tissue blocks were processed for paraffin embedding and subsequent staining by hematoxylin and eosin.

proximal and distal resection margins, uninvolved colon, and macroscopically abnormalities if any, such as polyps or other morphological changes. The mesenteric fat was thoroughly examined for lymph nodes. The obtained tissue blocks were processed for paraffin embedding and subsequent staining by hematoxylin and eosin. The presence of AC, MAC, or SC was documented. For demonstrating mucin, Periodic acid-Schiff staining was performed. Classical gland-forming adenocarcinomas with variable size and configuration of the glandular structures were classified as AC (ICD-O[International Classification of Diseases for Oncology]: 8140/3).2 MAC were defined as tumors with more than 50% of the lesion being composed of mucin, typically characterized by pools of extracellular mucin that contain malignant epithelium as acinar structures, strips of cells, or single cells (ICD-O: 8480/3).2 SC were defined as tumors with more than 50% of the lesion being composed of tumor cells with prominent intracytoplasmic mucin, typically characterized by large mucin vacuoles that fill the cytoplasm and displace the nucleus (ICD-O: 8490/3).2 According to the World Health Organization, all tumors were graded as well differentiated (G1), moderately differentiated (G2), poorly differentiated (G3), or undifferentiated (G4).2 Grading was determined by the content and appearance of glandular structures, without a priori differentiation between AC, MAC, and SC.2,16 Lymphatic invasion and angioinvasion was reported on the basis of hematoxylin and eosin staining.

rately differentiated (G2), poorly differentiated (G3), or undifferentiated (G4).2 Grading was determined by the content and appearance of glandular structures, without a priori differentiation between AC, MAC, and SC.2,16 Lymphatic invasion and angioinvasion was reported on the basis of hematoxylin and eosin staining. Statistical Analysis Statistical evaluation was performed using IBM SPSS statistics Version 19 (SPSS Inc; IBM Corporation Software Group, Somers, NY). The distribution of nominal- or ordinal-scaled variables was compared using Pearson χ2 test. Cardinal variables were tested for normal distribution by the Kolmogorov-Smirnov test. Explorative comparison of independent groups was performed by the t test for normal distribution and the Mann-Whitney U test (2 groups) or the Kruskal-Wallis test (more than 2 groups) for nonnormal distribution. All statistical tests were performed 2-sided, and P values less than 0.05 were considered to be statistically significant. No correction of P values was applied to adjust for multiple test issue. However, results of all statistical tests being conducted were thoroughly reported so that an informal adjustment of P values can be performed while reviewing the data.17 Multivariable analysis of binary outcome data was assessed by logistic regression. Time-dependent survival probabilities were estimated with the Kaplan-Meier method, and the log-rank (Mantel-Cox) test was used to compare independent subgroups. Cause-specific survival was used as the primary outcome parameter. Cause-specific survival is equivalent to disease-specific survival regarding the initial malignant disease and considers only tumor-related deaths as events.18 It reflects the intrinsic biology of the initial colorectal cancer more precisely than, for example, cancer-specific survival, which could be considered as death because of any kind of cancer. Cox proportional hazard models were used to investigate the effect on survival of multivariable relationships among covariates. Cause-specific survival times and estimated hazard ratios (HRs) were calculated and reported in 95% confidence intervals (95% CIs).

could be considered as death because of any kind of cancer. Cox proportional hazard models were used to investigate the effect on survival of multivariable relationships among covariates. Cause-specific survival times and estimated hazard ratios (HRs) were calculated and reported in 95% confidence intervals (95% CIs). RESULTS Patient Cohort Between January 1982 and October 2012, a total of 3881 patients underwent surgery for colorectal cancer at our institution. Of those patients, 402 were excluded because of nonresection procedures (defunctioning stoma/bypass) or histological subtypes other than AC, MAC, and SC. Thus, 3479 patients remained for analysis. Of these 3479 patients, 3074 had AC (88%), 375 had MAC (11%), and 30 had SC (0.9%). The median age was 65 years (range: 15–96 years). There were more men (n = 2016) than women (n = 1463) in the cohort. Intestinal tumor obstruction was present at the time of admission in 148 patients (4.3%). The tumor was located within the right hemicolon in 1052 patients (30%), within the left hemicolon in 991 patients (28%), and in the rectum in 1348 patients (39%). In 88 patients (2.5%), more than 1 colorectal cancer was found simultaneously. The prevalence of MAC and SC was not significantly altered after neoadjuvant treatment (P = 0.200). Of all 1348 patients with rectal cancer, 434 underwent neoadjuvant treatment. These 434 patients had AC in 90% (n = 389), MAC in 9% (n = 40), and SC in 1.2% (n = 5). On the contrary, 914 patients with rectal cancer did not undergo neoadjuvant treatment. Those had AC in 92% (n = 845), MAC in 7% (n = 63), and SC in 0.7% (n = 6).

348 patients with rectal cancer, 434 underwent neoadjuvant treatment. These 434 patients had AC in 90% (n = 389), MAC in 9% (n = 40), and SC in 1.2% (n = 5). On the contrary, 914 patients with rectal cancer did not undergo neoadjuvant treatment. Those had AC in 92% (n = 845), MAC in 7% (n = 63), and SC in 0.7% (n = 6). All patients underwent oncological resection of their primary tumor, which required multivisceral surgical procedures in 673 patients (19%). The median number of resected lymph nodes per patient was 16. UICC tumor stages were stage I in 805 patients (23%), stage II in 966 patients (28%), stage III in 942 patients (27%), and stage IV in 766 patients (22%). For all 2713 patients without metastasis (stages I–III), R0 resection status was achieved in 96% (n = 2612). For the 766 patients who presented with distant organ metastasis or peritoneal carcinomatosis (stage IV), an R0 resection was achieved in 19% (n = 148). Further characteristics of the patient cohort are shown in Table 1.

713 patients without metastasis (stages I–III), R0 resection status was achieved in 96% (n = 2612). For the 766 patients who presented with distant organ metastasis or peritoneal carcinomatosis (stage IV), an R0 resection was achieved in 19% (n = 148). Further characteristics of the patient cohort are shown in Table 1. TABLE 1. Clinicopathological Characteristics of the Patient Cohort The median follow-up time was 98 months (range: 1–311 months). During follow-up, 1347 patients (39%) had developed at least 1 recurrence (175 local recurrences, 1029 distant recurrences, 143 both local and distant recurrences). Data on recurrence were not available for 113 patients (3.2%). Postoperatively, 117 patients (3.4%) died within 30 days of operation. At the time of follow-up, 1629 patients (47%) were alive, 1154 patients (33%) had died from cancer-related causes, and 479 patients (14%) had died because of other causes. Survival data were not available for 100 patients (2.9%).

13 patients (3.2%). Postoperatively, 117 patients (3.4%) died within 30 days of operation. At the time of follow-up, 1629 patients (47%) were alive, 1154 patients (33%) had died from cancer-related causes, and 479 patients (14%) had died because of other causes. Survival data were not available for 100 patients (2.9%). Cancer Subtypes and Clinicopathology Age did not differ significantly between patients with AC, MAC, and SC. There was a higher proportion of women in the MAC group (47%, P = 0.034) than in the AC group (41%). Multivisceral surgery was performed more frequently for patients with MAC than for patients with AC (29% vs 18%, P < 0.001). Compared with AC (28%), MAC (51%, P < 0.001) and SC (50%, P = 0.029) were located more often within the right hemicolon. Furthermore, MAC and SC were diagnosed at more advanced stages than AC (P < 0.001 for each). UICC tumor stages III or IV were seen in 48% of patients with AC, in 55% of patients with MAC, and in 87% of patients with SC, respectively. Poor differentiation (G3/4) was found in 30% of patients with AC, in 45% of patients with MAC (P < 0.001, compared with AC), and in 90% of patients with SC (P < 0.001, compared with AC). Angioinvasion occurred more frequently in patients with AC than in patients with MAC (9% vs 5%, P = 0.011). Lymphatic vessel invasion occurred more frequently in patients with SC than in patients with AC (67% vs 25%, P < 0.001). For further details, see Table 2. Data regarding microsatellite instability were available for a subgroup of 251 patients, which reflected the relative frequencies of the histological subtypes in the whole collective. AC were microsatellite instable in 23% (50 of 221 patients), whereas MAC and SC were microsatellite instable in 50% (MAC: 14 of 28 patients, SC: 1 of 2 patients; P = 0.006).

lity were available for a subgroup of 251 patients, which reflected the relative frequencies of the histological subtypes in the whole collective. AC were microsatellite instable in 23% (50 of 221 patients), whereas MAC and SC were microsatellite instable in 50% (MAC: 14 of 28 patients, SC: 1 of 2 patients; P = 0.006). TABLE 2. Comparison of the Clinicopathological Characteristics of Patients With Adenocarcinoma, Mucinous Adenocarcinoma, and Signet-Ring Cell Carcinoma Survival Patients with MAC were more likely to develop local recurrence than those with AC (13% vs 8.6%, P = 0.014), whereas patients with SC were more likely to develop distant recurrence than those with AC (63% vs 33%, P = 0.002). Compared with the AC group (32%), more cancer-related deaths occurred in the MAC group (37%, P = 0.005) and SC group (77%, P < 0.001; Table 2). Cause-specific 5-year survival was 67 ± 1% for patients with AC, 61 ± 3% for patients with MAC, and 21 ± 8% for patients with SC (P < 0.001). Median cause-specific survival was not reached for AC and MAC. For SC, median cause-specific survival was 10 months (95% CI, 2–19 months). Compared with AC, cause-specific survival was significantly reduced for MAC (P = 0.011) and SC (P < 0.001). Furthermore, cause-specific survival was worse for SC compared with MAC (P < 0.001).

erence in survival between MAC and AC was no longer significant (HR = 1.0, 95% CI = 0.83–1.2; P = 0.991; Table 3). However, SC remained an independent predictor of poor prognosis, with an HR of 2.5 (95% CI = 1.6–3.8; P < 0.001). The stage-corrected survival characteristics confirmed these findings (Table 4 and Fig. 3). TABLE 3. Independent Prognostic Factors in Multivariable Analysis TABLE 4. Cause-Specific 5-Year Survival for Patients With Cancer With UICC Stages I, II, III, and IV, Corresponding to Adenocarcinoma, Mucinous Adenocarcinoma, and Signet-Ring Cell Carcinoma FIGURE 3. Cause-specific survival for patients of (A) stage I, (B) stage II, (C) stage III, and (D) stage IV, corresponding to adenocarcinoma, mucinous adenocarcinoma, and signet-ring cell carcinoma. Levels of significance are listed in Table 4. DISCUSSION In this study, MAC and SC were shown to be uncommon, respectively, rare histological colorectal cancer subtypes with differences in clinical and histopathological findings. Compared with patients with AC, patients with MAC had more advanced tumor stages and reduced survival. SC were rare; however, these tumors implicated a clearly different behavior and reduced survival even after correction for tumor stage.

ctal cancer subtypes with differences in clinical and histopathological findings. Compared with patients with AC, patients with MAC had more advanced tumor stages and reduced survival. SC were rare; however, these tumors implicated a clearly different behavior and reduced survival even after correction for tumor stage. To our knowledge, we report on the largest European cohort investigating patients with AC, MAC, and SC and confirm previous nationwide epidemiologic data from the United States and Asia.9,10,19 The number of patients we studied is considerably smaller than in population-based reports; nevertheless, our thoroughly documented clinicopathological and long-term follow-up data are a strength of the cohort we describe. The prevalence of 11% for MAC and 0.9% for SC in our cohort is in line with population-based data sets.9,10,19 Patients suffering from these tumor subtypes may be candidates for individually tailored treatment regimens, such as intensified systemic therapy and more frequent follow-up. Nevertheless, no prospective study has yet been performed to investigate different management of these entities, and specific therapeutic regimens are not recommended in current guidelines.7

may be candidates for individually tailored treatment regimens, such as intensified systemic therapy and more frequent follow-up. Nevertheless, no prospective study has yet been performed to investigate different management of these entities, and specific therapeutic regimens are not recommended in current guidelines.7 In accordance with the majority of published series including cohorts from the United States and Asia, MAC and SC were largely located within the right hemicolon,4,9,10,20,21 had higher stages of the primary tumor4,5,13,20 and lymph node involvement,5,13,14,20,21 and more advanced UICC stages4,10,13,21 and poorer differentiation.4,9,10,14 Differences in these variables were more pronounced between SC and AC than between MAC and AC. AC of the right hemicolon were not associated with higher tumor stages than those of the left hemicolon. Thus, a putative delayed diagnosis for right-sided colon cancer in general does not explain the advanced tumor stages of MAC and SC. The more aggressive tumor biology of MAC and particularly SC is more likely.

d AC. AC of the right hemicolon were not associated with higher tumor stages than those of the left hemicolon. Thus, a putative delayed diagnosis for right-sided colon cancer in general does not explain the advanced tumor stages of MAC and SC. The more aggressive tumor biology of MAC and particularly SC is more likely. Against our expectations, patients with MAC had low rates of angioinvasion but more local recurrences, whereas patients with SC had high rates of lymphatic invasion and more metachronous distant metastases. However, these results should be interpreted carefully because angioinvasion and lymphatic invasion were reported on the basis of routinely performed hematoxylin and eosin staining. A reduced sensitivity compared with detailed immunohistochemical staining methods cannot be excluded. More complete tumor resections (R0) were performed for MAC than for SC. However, the higher number of local recurrences for MAC could be due to higher rates of multivisceral resections in this group. The 3 large population-based studies from the United States9,10 and Asia19 did not investigate angioinvasion or lymphatic invasion, and there is no study that confirms or contradicts our findings. Thus, generalizability may be restricted because of the relatively small number of patients in our cohort. Overall, only a few studies have reported on angioinvasion and lymphatic invasion of AC, MAC, and SC so far.4,6,20,21 Lee et al21 and Sung et al6 found higher rates of angioinvasion in SC than in MAC; however, they did not evaluate AC. The same observation was made in our cohort when comparing SC with MAC (data not shown).

all, only a few studies have reported on angioinvasion and lymphatic invasion of AC, MAC, and SC so far.4,6,20,21 Lee et al21 and Sung et al6 found higher rates of angioinvasion in SC than in MAC; however, they did not evaluate AC. The same observation was made in our cohort when comparing SC with MAC (data not shown). There was a significant shorter cause-specific survival for MAC than that for AC and for SC than that for AC and MAC. However, multivariable analysis including TNM stage confirmed SC to be an independent prognostic factor but not MAC. Survival curves confirmed no significant differences between MAC and AC when analyzed by tumor stage (Table 4 and Fig. 3). These findings are reflected in the current literature.4,5,8–10,14,20,21 However, small reports6,13 and 2 meta-analyses3,12 identified MAC as an independent prognostic factor for poor outcome. Study heterogeneity of the prognostic impact of MAC was high between the studies included in the meta-analyses, and the pooled HR did not exceed 1.05 (95% CI = 1.02–1.08, P < 0.001)3 and 1.06 (95% CI = 1.04–1.08, P < 0.001),12 respectively. Therefore, a significant prognostic effect of MAC is possible but might not be clinically relevant. In our multivariable analysis, the HR for MAC compared with AC was 1.001 (95% CI = 0.83–1.21, P = 0.991).

HR did not exceed 1.05 (95% CI = 1.02–1.08, P < 0.001)3 and 1.06 (95% CI = 1.04–1.08, P < 0.001),12 respectively. Therefore, a significant prognostic effect of MAC is possible but might not be clinically relevant. In our multivariable analysis, the HR for MAC compared with AC was 1.001 (95% CI = 0.83–1.21, P = 0.991). Neoadjuvant treatment is known to produce mucin pools.22 However, the prevalence of MAC or SC was not significantly increased after neoadjuvant treatment. Furthermore, survival of patients with MAC who underwent neoadjuvant chemoradiation did not significantly differ from those without preoperative treatment. Resistance to chemoradiation for MAC has been suggested recently.3 However, neither data from our study nor data from the current literature provide compelling evidence.3 For patients with MAC, late diagnosis at more advanced tumor stages and the observed high risk of local recurrence may be clinically more important than a slight difference in survival.

has been suggested recently.3 However, neither data from our study nor data from the current literature provide compelling evidence.3 For patients with MAC, late diagnosis at more advanced tumor stages and the observed high risk of local recurrence may be clinically more important than a slight difference in survival. Regarding tumor biology, AC, MAC, and SC do not seem to represent consecutive steps of de-differentiation during the development of mucus-producing colorectal carcinoma but rather constitute different genetic pathways.5,11 Our study revealed distinct histopathological features and recurrence patterns for MAC and SC (Table 2), which were discussed previously. These results are supported by described genetic profiles specific for MAC and SC, including microsatellite instability and mutations of the genes BRAF, p53, and p16.4,5 In addition, downregulation of the adhesion molecules E-cadherin and β-catenin may lead to reduced cell adhesion in areas of high mucus content and promote scattering of tumor cells, further leading to advanced tumor stages and poorer prognosis.5,6 In our cohort, high rates of local recurrences including localized peritoneal carcinomatosis for patients with MAC and distant recurrences for patients with SC might reflect the aggressive phenotype of MAC and SC. In particular, local invasiveness may be increased in areas of high extracellular mucus content, as it is found in MAC.

, high rates of local recurrences including localized peritoneal carcinomatosis for patients with MAC and distant recurrences for patients with SC might reflect the aggressive phenotype of MAC and SC. In particular, local invasiveness may be increased in areas of high extracellular mucus content, as it is found in MAC. SC have been described as being positive for intestinal trefoil factor and MUC2, 2 peptides that are usually produced only by goblet cells.5 Thus, SC could arise from different cells of origin than AC.5,23 Although they can be localized in the colorectum, SC may be genetically more related to signet-ring cell cancers of other organs (eg, gastric cancer) than to AC or MAC of the colorectum.5,11 The absence of E-cadherin/β-catenin and amplification of Bcl-2 are features typically shared with signet-ring cell cancer of the stomach but not with classical colorectal adenocarcinomas.5

genetically more related to signet-ring cell cancers of other organs (eg, gastric cancer) than to AC or MAC of the colorectum.5,11 The absence of E-cadherin/β-catenin and amplification of Bcl-2 are features typically shared with signet-ring cell cancer of the stomach but not with classical colorectal adenocarcinomas.5 In the past, all MAC and SC were classified as “poorly differentiated/high grade.”16 However, since 2010, the World Health Organization classification2 noted that grade of epithelial differentiation indeed formally determines the pathological grading (G1–G4, also described for the patients reported here). Microsatellite instability is an important molecular marker that occurs more frequently in MAC and SC than in classical AC and allows for risk stratification.2,7,16 Because of the association with right-sided tumors, the increased rate of microsatellite instability in MAC and SC may be caused by its embryologic origin. Although the distal part of the colon derives from the hindgut, the proximal part derives from the midgut and exhibits specific cellular genetic characteristics.24 Microsatellite instable tumors are associated with better prognosis and are currently classified as “low-grade.” If MAC or SC are microsatellite stable, they are associated with more aggressive biological behavior and are classified as “high-grade.”2,16,25 In current practice, microsatellite testing should be performed for all patients with MAC and SC.16 Because of lack of tissue samples from early patients of this cohort, we do not report on microsatellite data in detail here. In accordance with other reports,6 higher rates of microsatellite instability were detected in MAC and SC than in AC in a subgroup analysis (P = 0.006). Patients with microsatellite instable tumors had a better survival25; however, these findings were not statistically significant, most likely due to the small size of the individual groups (data not shown).

er rates of microsatellite instability were detected in MAC and SC than in AC in a subgroup analysis (P = 0.006). Patients with microsatellite instable tumors had a better survival25; however, these findings were not statistically significant, most likely due to the small size of the individual groups (data not shown). CONCLUSIONS MAC and SC are uncommon, respectively, rare subtypes of colorectal cancer with a worse prognosis than classical AC. They are considered as independent tumor entities with different biological behavior. Our data suggest that patients with MAC and SC could profit from closer follow-up or even intensified adjuvant therapy because of their high rates of local and distant recurrence. The biological behavior of SC differs in specific, and these patients require special awareness, despite the relatively rare prevalence. Reprints: Helmut Friess, MD, Hötzlring 22, 81737 Munich, Germany. E-mail: helmut@friess.cc. Disclosure: Supported by a grant from the KKF (Kommission für klinische Forschung, Fakultät für Medizin, Technische Universität München) to U.N. and K.-P.J. The authors declare no conflicts of interest. DISCUSSANTS E. Rullier (Bordeaux, France): I want to congratulate you for this very impressive cohort of more than 3000 patients extracted from a prospective database of colorectal cancers operated on at a single institution.

Disclosure: Supported by a grant from the KKF (Kommission für klinische Forschung, Fakultät für Medizin, Technische Universität München) to U.N. and K.-P.J. The authors declare no conflicts of interest. DISCUSSANTS E. Rullier (Bordeaux, France): I want to congratulate you for this very impressive cohort of more than 3000 patients extracted from a prospective database of colorectal cancers operated on at a single institution. My first question concerns the period of evaluation that is very long, 30 years, and you know that during this period the TNM classification has been changed many times. So, how did you manage the pathologic classification of your database? Did you reclassify the tumors; did you review all the cases asking a single pathologist to do that? The second question is about the resection that is very impressive, 96% of complete microscopic resection. I suspect that multivisceral resection is the explanation for that but I also suspect that maybe you have some incomplete or missing data, because the concept of R0 resection came after the beginning of your study. So, my question is about the definition of R0 resection. Could you specify what was the definition, was it only longitudinal or also circumferential resection margin? Again, a question about the pathology methodology; you know that assessment of irradiated specimens, especially rectal resections, is very difficult because irradiation produces mucin pools. Did you have any difficulty in distinguishing between natural and irradiated mucin pools because you determined that 10% of your specimens were mucinous tumors?

hology methodology; you know that assessment of irradiated specimens, especially rectal resections, is very difficult because irradiation produces mucin pools. Did you have any difficulty in distinguishing between natural and irradiated mucin pools because you determined that 10% of your specimens were mucinous tumors? During your presentation, you did not present the local recurrence data, but I read your article and you used local recurrence to demonstrate the potential difference in terms of biology and prognoses in your subgroup of mucinous tumors. In your article, you conclude that local recurrence is higher in the group of mucinous tumors. I am not sure I agree with you on that because you demonstrate clearly that the groups are not the same. You showed that there is higher stage in the mucinous group than in the conventional one. So finally, when you analyzed survival, the worse prognoses of mucinous tumors disappeared after multivariate analysis. Did you use multivariate analysis also for local recurrence, which is not described in your article and not presented here? I suspect that the main conclusion of the article would be that signet-ring cell but not mucinous tumors may have a different biology and prognoses as compared with conventional colorectal cancer.

Did you use multivariate analysis also for local recurrence, which is not described in your article and not presented here? I suspect that the main conclusion of the article would be that signet-ring cell but not mucinous tumors may have a different biology and prognoses as compared with conventional colorectal cancer. Response From U. Nitsche (Munich, Germany): Thank you for these encouraging comments and questions. Regarding your first question, the long review period, you are correct. During these last 30 years, there were 5 different TNM classification systems, but, indeed, we did reclassify all the patients regarding the latest, seventh TNM classification system, which was possible by reviewing the pathological reports and our data. You also asked whether all the specimens were reviewed by a single pathologist; this was not the case. What we did was to review all the pathological reports and we included or defined only those tumors as mucinous or signet-ring carcinomas in which we clearly had the definition according to the pathological report or at least the statement of the amount of mucin pools. But you raise a good point because nearly every tumor is described as having some mucus component, but that does not qualify every tumor as a mucinous carcinoma. You mentioned the multivisceral resection and the high R0 rate. Indeed, we had a 96% R0 rate for stages 1 to 3 and this was possible by a 19% rate of multivisceral resection. However, for the stage 4 disease, we obtained R0 in 19%.

You also asked whether all the specimens were reviewed by a single pathologist; this was not the case. What we did was to review all the pathological reports and we included or defined only those tumors as mucinous or signet-ring carcinomas in which we clearly had the definition according to the pathological report or at least the statement of the amount of mucin pools. But you raise a good point because nearly every tumor is described as having some mucus component, but that does not qualify every tumor as a mucinous carcinoma. You mentioned the multivisceral resection and the high R0 rate. Indeed, we had a 96% R0 rate for stages 1 to 3 and this was possible by a 19% rate of multivisceral resection. However, for the stage 4 disease, we obtained R0 in 19%. With regard to the earlier pathological reports, I'm not quite sure when resection was first included in the TNM classification but, of course, whenever it was stated, we included it for our analysis. In the earlier cases when R0 may have not been stated, at least our pathologist routinely reported for every tumor if there were microscopic or macroscopic tumor cells at the resection margin, which obviously is one of the most important findings for the surgeon. This was done for all specimens as well for the longitudinal and also for the circumferential resection margin.

at least our pathologist routinely reported for every tumor if there were microscopic or macroscopic tumor cells at the resection margin, which obviously is one of the most important findings for the surgeon. This was done for all specimens as well for the longitudinal and also for the circumferential resection margin. Your next question was a very interesting one, the rate of mucinous adenocarcinomas or of cancers with mucinous differentiation, which obviously according to the literature is higher after neoadjuvant radiation. Unfortunately, we were not able to investigate whether there was a higher rate of mucinous adenocarcinomas after radiation as we did not have the preoperative findings, but we did investigate if there was a higher rate of mucinous adenocarcinomas for patients with rectal cancer if they were radiated in comparison to if they did not undergo preoperative radiation. And this was the case, it was 7% compared with 9%, which was not significant.

as we did not have the preoperative findings, but we did investigate if there was a higher rate of mucinous adenocarcinomas for patients with rectal cancer if they were radiated in comparison to if they did not undergo preoperative radiation. And this was the case, it was 7% compared with 9%, which was not significant. I think we also agree with your final comment. The mucinous adenocarcinomas were diagnosed at higher tumor stages and this may have contributed to the higher rates of local recurrence. We did a multivariable analysis on these data and found similar results to the other multivariable analysis; the mucinous adenocarcinoma was not an independent predictor of local recurrence. We concluded that mucinous adenocarcinomas do represent some specific characteristics that clearly differ from the classical adenocarcinomas. However, the signet-ring cell carcinomas seem really to represent another different histological or biological type. DISCUSSANTS N. Senninger (Munster, Germany): These findings are shown in a large series and you should be congratulated because we all know how difficult it is to bring back the files of previous surgical generations. The second point is not completely new because in pathology, in the 1970s we have been told about the difficult biological properties of tumors.

): These findings are shown in a large series and you should be congratulated because we all know how difficult it is to bring back the files of previous surgical generations. The second point is not completely new because in pathology, in the 1970s we have been told about the difficult biological properties of tumors. My question is, we saw the advent of neoadjuvant regimens, new drugs, and so forth in the last 30 years, and how could you make Kaplan-Meier curves with patients who had been treated so differently inside the last 30 years? If you did not do that, I would strongly recommend that you compare the decades, first decade, second, and third, because there the patients would be rather homogenous. Response From U. Nitsche (Munich, Germany): I think one of the strengths of our series, even if not all the findings were new, is that we had very complete documentation of lymphatic and vascular invasion; however, you are correct, we did not compare the Kaplan-Meier curves regarding decades. Perhaps that is a good idea. However, to our mind the treatment may have shifted during recent years. This should be the same effect for all cancers in regard to the histological subtypes. Therefore, we do not assume that a bias may be raised because of this factor.

ot compare the Kaplan-Meier curves regarding decades. Perhaps that is a good idea. However, to our mind the treatment may have shifted during recent years. This should be the same effect for all cancers in regard to the histological subtypes. Therefore, we do not assume that a bias may be raised because of this factor. DISCUSSANTS J. Izbicki (Hamburg, Germany): I enjoyed your presentation very much. Following on Professor Senninger's comment, it is not a new finding that a signet-ring cell carcinoma had a different biological behavior, worse prognosis than a normal adenocarcinoma. So, my at least a little bit provocative question is what is exactly new in your presentation? The second comment is with regard to the pathological workup. Did I understand this correctly that you just adhered to the pathological written reports and reviewed them or did a specific pathologist rereview the slides of the specimen? Response From U. Nitsche (Munich, Germany): We adhered to the pathological reports and we did not review every single slide of all the more than 3000 patients. And to your first question, indeed, there have been many reports that signet-ring cell cancers are different in terms of behavior. However, our study still remains a large analysis that compares the 3 most frequent cancer types; mucinous, classical adenocarcinoma, and signet-ring cell carcinomas. We report on comprehensive markers such as lymphatic invasion, lymph node invasion, local recurrence rate, or overall survival that has not been done in this extensive method so far.

ins a large analysis that compares the 3 most frequent cancer types; mucinous, classical adenocarcinoma, and signet-ring cell carcinomas. We report on comprehensive markers such as lymphatic invasion, lymph node invasion, local recurrence rate, or overall survival that has not been done in this extensive method so far. DISCUSSANTS P. Gertsch (Zurich, Switzerland): This is a very nice study and I congratulate you. Those treating peritoneal carcinomatosis often see the dismal prognoses of appendix tumors with signet-ring histology. In your presentation, you mentioned that most signet-ring cell carcinomas were on the right side of the colon. Have you been able to differentiate the appendix from the right colon tumor? Response From U. Nitsche (Munich, Germany): We included appendix tumors with the right-sided cancers but you make a good point, the appendix tumors can be different and include neuroendocrine tumors. If that was the case, they were excluded from this analysis as we investigated only the classical, the mucinous, and the signet-ring cell carcinomas.

Recent years have seen major changes in surgical training worldwide, with reduced working hours enforced under European law,1 and the introduction of run-through and core training programmes in the United Kingdom, which no longer expose the surgical trainee to all the common surgical conditions and operations that would have formed part of traditional surgical training. Trends to reduce working hours are also evident in North America2 and Australasia.3 In addition, there has been a lack of teaching provision of knowledge and consistent educational support as the trainee has approached the major compulsory examination of their early professional career, Membership examination of the Royal College of Surgeons (MRCS), normally taken during the first 2 years of basic surgical training. In response to these changes, a jointly awarded modular qualification, the Edinburgh Surgical Sciences Qualification (ESSQ—http://www.essq.rcsed.ac.uk), was launched in 2007 by the University of Edinburgh and the Royal College of Surgeons of Edinburgh (RCSEd). The programme was based on the MRCS curriculum but was tailored to be equally applicable to the early years of surgical training programmes worldwide.

al Sciences Qualification (ESSQ—http://www.essq.rcsed.ac.uk), was launched in 2007 by the University of Edinburgh and the Royal College of Surgeons of Edinburgh (RCSEd). The programme was based on the MRCS curriculum but was tailored to be equally applicable to the early years of surgical training programmes worldwide. The online ESSQ provides learning material tailored to the professional needs of surgical trainees, delivered in a flexible and readily accessible manner that enables the student to measure their progress through the individual components of the MRCS curriculum, while providing peer support and senior mentorship during this critical phase of their professional development. The online clinical case scenarios have specific objectives that underpin the curriculum in the applied surgical sciences of anatomy, radiology, physiology, pathology and bacteriology, with modules in the assessment and care of the surgical patient, research, and communication skills. The aim of this study was to determine the impact of the ESSQ e-learning programme in supporting the early academic development of surgical trainees in their preparation for professional examination. The success of trainees in completing their MSc degree and in securing their professional examination (MRCS) was assessed and the quality of the programme determined by securing direct student feedback.

in supporting the early academic development of surgical trainees in their preparation for professional examination. The success of trainees in completing their MSc degree and in securing their professional examination (MRCS) was assessed and the quality of the programme determined by securing direct student feedback. METHODS The ESSQ programme is intended for doctors in the early stages of their careers (pre-MRCS), and therefore those who already possessed the MRCS or Fellowship of the Royal College of Surgeons (FRCS) were ineligible to enroll. Applications were accepted only from trainees in possession of a medical degree from their country of origin (MBChB or equivalent) and who were based within a training environment. The programme provides academic recognition of trainees' acquisition of knowledge and development; the ESSQ leads progressively to a Certificate (year 1—basic science relevant to surgery), Diploma (year 2—clinical surgery) and an MSc in Surgical Sciences (year 3—research component) over a 3-year period of part-time study that runs alongside and complements full-time clinical training. Students in the first 2 years of the programme have access to e-tutor–assessed collaborative discussion boards, where they are expected to lead debate and facilitate interaction with others and bring original material to support their arguments (http://demo.essq.ed.ac.uk/Surgical_Oncology_Discussion.asp), in addition to completion of multiple-choice questions (MCQs) and mini essays, with formative feedback provided. Summative assessment of knowledge is undertaken through an end-of-year examination comprising single best answer questions and extended matching item questions in a similar format to the MRCS examination. In addition, there is an in-course assessment score reflecting the student's performance in various aspects of the course throughout the year. During the second year, students are also introduced to the basic concepts relating to their academic development in preparation for the final year, in which students are afforded the opportunity to undertake a period of novel research leading to a dissertation and the award of MSc in surgical sciences. Although the primary objective of the third year is to consolidate academic principles and learning, many individuals go on to publish their work in peer-reviewed journals.

students are afforded the opportunity to undertake a period of novel research leading to a dissertation and the award of MSc in surgical sciences. Although the primary objective of the third year is to consolidate academic principles and learning, many individuals go on to publish their work in peer-reviewed journals. Academic data regarding performance and assessment (end-of-year scores) were sourced from student records relating to academic years 2007–2008 to 2011–2012 of the MSc in Surgical Sciences programme at the University of Edinburgh, in partnership with the Royal College of Surgeons of Edinburgh. Demographic data collected included age, gender, and the training post held at the time of entry to the programme. In years 1 and 2, students were assessed on the basis of completion of in-course MCQ, discussion board contributions, essays (year 2 only), and final examination marks. The relative weighting of in-course assessment increased over the 3 years: 25% in year 1, 30% in year 2, and 50% in year 3. In year 3, students completed a research project proposal and summary, an e-poster, and a final e-dissertation, all of which contributed to the end-of-year score. All examination questions were quality assured (including Angoff standard setting) and the examination and in-course assessment results were ratified by a programme Board of Examiners at the University of Edinburgh, which included 2 external examiners, both consultant surgeons with academic appointments at other UK institutions, which included responsibilities for postgraduate clinical teaching. The selection of external examiners and conduct of the board of examiners were done in accordance with the University's Taught Assessment Regulations and Guidelines.

aminers, both consultant surgeons with academic appointments at other UK institutions, which included responsibilities for postgraduate clinical teaching. The selection of external examiners and conduct of the board of examiners were done in accordance with the University's Taught Assessment Regulations and Guidelines. Anonymized student feedback was examined over the same period using archived responses to online questionnaires. Such feedback was sought on agreed pro forma that were circulated to students at the end of each academic year. A search of PubMed and Web of Knowledge was undertaken using ESSQ student name, supervisor name, and research topic to identify published research articles and abstracts resulting from the dissertation topic.

Such feedback was sought on agreed pro forma that were circulated to students at the end of each academic year. A search of PubMed and Web of Knowledge was undertaken using ESSQ student name, supervisor name, and research topic to identify published research articles and abstracts resulting from the dissertation topic. The MRCS is designed especially for surgical trainees in Great Britian and Ireland and is normally taken during the first 24 months of surgical training. Part A of the MRCS is a written examination on generic surgical sciences and applied knowledge, whereas the Part B Objective Structured Clinical Examination integrates basic surgical scientific knowledge and its application to clinical surgery. An e-mail was sent to all students who had completed both years 1 and 2 of the ESSQ, requesting their consent for the programme team to obtain MRCS Part A examination scores. Where consent was given, MRCS results were obtained from the “Assessment and Quality Assurance Examinations” section of the RCSEd. In addition, pass rates (of 100%) were obtained from the RCSEd for all MRCS examinations held during 2008 and 2012. Data were separated into pass rates for ESSQ students and non-ESSQ candidates. Immediately after their correlation with performance in the MSc programme, all data relating to individuals' MRCS examination scores were anonymized.

(of 100%) were obtained from the RCSEd for all MRCS examinations held during 2008 and 2012. Data were separated into pass rates for ESSQ students and non-ESSQ candidates. Immediately after their correlation with performance in the MSc programme, all data relating to individuals' MRCS examination scores were anonymized. Statistical analysis was performed using SPSS Statistics 19 and GraphPad Prism 3.0 software. Differences between group means were identified using Student t test for 2-group comparisons, and 1-way analysis of variance for multiple group comparisons. The influence of age, gender, and level at entry on end-of-year scores was assessed using the Cox proportional hazards regression model. Linear regression analysis was carried out to determine the relationship between ESSQ performance and MRCS examination scores. Data are expressed as mean ± SD unless otherwise stated. Statistical significance was accepted when P < 0.05. RESULTS A total of 517 surgical trainees from 40 countries (Fig. 1) were admitted to the programme over the 6-year (2007–2012) period observed. Of the 205 trainees who entered the 3-year programme during 2007–2009, 116 completed successfully and have graduated with an MSc degree (57%), 36 (18%) completed at Diploma level, 26 (13%) completed at Certificate level, 4 (2%) have returned from an interruption of study, and 23 (11%) withdrew early with no award. The recruitment and overall progression rates of the ESSQ programme over the 5 completed academic years are shown in Table 1.

an MSc degree (57%), 36 (18%) completed at Diploma level, 26 (13%) completed at Certificate level, 4 (2%) have returned from an interruption of study, and 23 (11%) withdrew early with no award. The recruitment and overall progression rates of the ESSQ programme over the 5 completed academic years are shown in Table 1. FIGURE 1. The online distance-learning ESSQ/MSc in Surgical Sciences attracts students from 40 different countries. TABLE 1. ESSQ Recruitment and Progression Data The majority of students, 279 of the 368 (76%) whose level of work experience was known, entered as foundation doctors (interns) within 2 years of graduation from medical school and before commencing formal surgical training. Of the remaining 89 trainees, 68, 18, and 3 students were in the first, second, and third years, respectively, of a surgical training programme (Fig. 2). Level of work experience at entry to programme was not found to influence performance across each of the 3 years of the programme (Fig. 3). On multivariate analysis, age, gender, and level of work experience at entry were not found to be predictors of end-of-year scores (Table 2). FIGURE 2. Surgical training structure in the United Kingdom. The trainee has to demonstrate the acquisition of knowledge and technical skills in the workplace and negotiate examinations of the Royal College of Surgeons (Membership, MRCS, and Fellowship, FRCS), which are mandatory requirements in demonstrating competency in surgery.

gical training structure in the United Kingdom. The trainee has to demonstrate the acquisition of knowledge and technical skills in the workplace and negotiate examinations of the Royal College of Surgeons (Membership, MRCS, and Fellowship, FRCS), which are mandatory requirements in demonstrating competency in surgery. FIGURE 3. The influence of level at entry to the ESSQ programme and academic performance (2007–2012). Individual marks (with mean values) shown for students at the end of (A) year 1 (n = 279), (B) year 2 (n = 188), and (C) year 3 (n = 117) reveal that there is no significant effect of level of training upon entering the programme and end-of-year results (P = 0.26, 0.12, and 0.48 for years 1, 2, and 3, respectively; 1-way analysis of variance). FY1/2 = Foundation year doctors (year 1 or 2 after graduation); ST1/2 = first-/second-year surgical trainee, equivalent to first-/second-year core trainee (CT1/2); ST3 = third-year surgical trainee.

ramme and end-of-year results (P = 0.26, 0.12, and 0.48 for years 1, 2, and 3, respectively; 1-way analysis of variance). FY1/2 = Foundation year doctors (year 1 or 2 after graduation); ST1/2 = first-/second-year surgical trainee, equivalent to first-/second-year core trainee (CT1/2); ST3 = third-year surgical trainee. TABLE 2. Multivariate Analysis of the Influence of Gender, Age, and Training Level at Entry, on Academic Performance on the ESSQ Programme Average pass rates since the programme launched were 84% ± 11% in year 1, 85% ± 10% in year 2, and 88% ± 7% in year 3. MSc students had significantly higher MRCS pass rates than nonenrolled trainees (Fig. 4), for both parts A and B of the examination. Comparison of the groups sitting the MRCS examinations revealed that candidates were similar in terms of age and expected level of clinical experience: ESSQ and non-ESSQ candidates both had an average age of 26 years. Furthermore, there was a significant correlation between MRCS examination performance and overall performance in the first 2 years of the MSc programme (Fig. 5). FIGURE 4. Pass rate in 2008-2012 for students enrolled (ESSQ) or not enrolled (non-ESSQ) in the ESSQ. (A) Combined or overall examination pass rate (mean ± SD; *P < 0.01, n = 352; unpaired student t test) and (B) separate part A and B examination pass rate (mean ± SD; *P < 0.05, n = 223; †P < 0.01, n = 129; unpaired Student t test).

rate in 2008-2012 for students enrolled (ESSQ) or not enrolled (non-ESSQ) in the ESSQ. (A) Combined or overall examination pass rate (mean ± SD; *P < 0.01, n = 352; unpaired student t test) and (B) separate part A and B examination pass rate (mean ± SD; *P < 0.05, n = 223; †P < 0.01, n = 129; unpaired Student t test). FIGURE 5. Regression analysis showing the association between student performance in the first 2 years of the ESSQ and MRCS part A examination outcome (y = 0.6505x + 34.859; R2 = 58%; P < 0.01, n = 37) (95% confidence intervals). Feedback from online, anonymous questionnaires revealed that 224 (90%) of 248 respondents would recommend the programme to their peers and the same percentage reported that they had a better understanding of the basic and applied surgical sciences than they would have otherwise had if they had not enrolled on the ESSQ programme. The majority of students (81%) considered that the ESSQ would improve their chances of gaining a surgical training post. In addition, 61% found the level of the MCQs within the ESSQ programme to be more difficult than those in the MRCS examination. On average, year 1 students felt that they needed to spend 10 ± 6 hours per week engaged in ESSQ activities to keep up-to-date with the timetable. Respondents consistently reported that they found the ESSQ an enjoyable experience. Search of PubMed and Web of Knowledge revealed that there have been 26 full research papers and 27 abstracts published to date relating to ESSQ research dissertations submitted May 2010 onwards.

es to keep up-to-date with the timetable. Respondents consistently reported that they found the ESSQ an enjoyable experience. Search of PubMed and Web of Knowledge revealed that there have been 26 full research papers and 27 abstracts published to date relating to ESSQ research dissertations submitted May 2010 onwards. DISCUSSION The ESSQ/MSc in Surgical Sciences was established in 2007 as a response to changes in employment law in Europe, which reduced clinical exposure in the workplace.1 Contemporaneously, changes in surgical training were implemented with the aim of streamlining progression through training milestones and shortening the duration of training. The entire ESSQ programme, based on the MRCS curriculum, is taught through self-directed learning at a distance. It is delivered via a virtual learning environment which, allows students easy access to internal and external coursework, on an “anytime, anywhere” basis via the Internet on computer or mobile devices. The learning style provides students the options of a problem-oriented entry or systematic learning illustrated through subsequent problem solving, delivered in modules scheduled in blocks of time that accommodate and complement the student's clinical training demands.

is via the Internet on computer or mobile devices. The learning style provides students the options of a problem-oriented entry or systematic learning illustrated through subsequent problem solving, delivered in modules scheduled in blocks of time that accommodate and complement the student's clinical training demands. Since its inception, the ESSQ has consistently exceeded its recruitment target with the highest student numbers of any postgraduate programme at the University of Edinburgh. Although targeted specifically at surgical trainees based in the United Kingdom, it has attracted more than 500 students from 40 countries in the last 6 years. The programme appears, therefore, to have a growing international reach with applicability to the early years of surgical training in many countries. This may be related to the current general reduction in working experience for doctors; the requirement to reduce working hours in Europe, North America, and Australasia1–3 has diminished considerably the trainee's exposure to both common and less frequently encountered surgical conditions.4,5 There is, therefore, a need to consolidate the applied knowledge that may be compromised by reduced exposure to the assessment of the patient in both the elective and emergency setting.6,7 We view distance-learning surgical educational programmes as an approach different from and complementary to traditional vocational methods, and which serve to enhance the learning experience of surgeons in training, not to replace but to complement experience in the workplace. The high recruitment to the programme and positive feedback from students would endorse this view.

as an approach different from and complementary to traditional vocational methods, and which serve to enhance the learning experience of surgeons in training, not to replace but to complement experience in the workplace. The high recruitment to the programme and positive feedback from students would endorse this view. The design of the ESSQ programme is underpinned by the pedagogical principles of e-learning, whereby students are supported as members of an online community with clearly defined learning objectives and use of materials that are active, engaging, collaborative, and relevant.8,9 Ease of navigation around the virtual learning environment and regular support from the programme team are critical in achieving successful online learning.10 The e-learning platform developed in-house (eeSURG, http://learning.essq.ed.ac.uk) allows students to navigate resources easily, follow the programme curriculum, access course material, and interact with course tutors and other students. Central to course design, Blanchard and Frasson11 emphasize the importance of roleplay to enhance constructivist learning. The activity model within the eeSURG platform is provided by Labyrinth (http://labyrinth.mvm.ed.ac.uk), a tool for authoring and delivering case narratives. Labyrinth was designed in-house with the goal of supporting development of richly engaging, narrative medical cases that invite users to take control of their own decisions—and, by extension, their own learning—and develop the critical analysis skills to face effectively the consequences of those decisions.12 It is noted that there was an appreciable “dropout” rate from the programme each year (10%–15%) but this was less than that observed in other distance learning masters programmes; online courses are associated with higher attrition rates than on-campus equivalents,13 with several studies reporting dropout rates for online masters programmes exceeding 40%.14

le “dropout” rate from the programme each year (10%–15%) but this was less than that observed in other distance learning masters programmes; online courses are associated with higher attrition rates than on-campus equivalents,13 with several studies reporting dropout rates for online masters programmes exceeding 40%.14 Although the online ESSQ/MSc in Surgical Sciences is designed to be studied part-time, alongside full-time clinical training in a hospital setting, the absence of a significant association between students' training level and academic performance on the ESSQ programme suggests that the content of the programme stands alone and is not necessarily dependent on previous surgical experience. Greater experience held by more advanced trainees may be offset by a longer period away from formal academic study, conferring an advantage to recent graduates who are more accustomed to intensive study. Furthermore, it is recognized that each student within a particular level of foundation (intern) or surgical training can have a very different profile of skills and experience compared with others at the same level, accounting for the lack of a relationship between performance and level at entry to the programme.

udy. Furthermore, it is recognized that each student within a particular level of foundation (intern) or surgical training can have a very different profile of skills and experience compared with others at the same level, accounting for the lack of a relationship between performance and level at entry to the programme. Given that one of the principal aims of the ESSQ programme was to prepare the surgical trainee better for the individual components of the MRCS examination that were covered in the first 2 years to Diploma level, monitoring pass rates for those students who have presented themselves for MRCS examination in the United Kingdom has provided a valuable assessment of outcome. The increased average pass rate for ESSQ students (∼15% higher) than non-ESSQ students add credence to the positive appraisal of the programme to date and will contribute to future student engagement and motivation. The correlation between performance in the ESSQ programme and MRCS examination scores and the positive feedback from students further demonstrates that the programme is fit-for-purpose. It is accepted that a potential limitation of the current analysis is that the ESSQ may not improve MRCS outcomes per se, but rather the ESSQ students are a self-selected group of highly motivated students who were more likely to be successful in the MRCS.

nts further demonstrates that the programme is fit-for-purpose. It is accepted that a potential limitation of the current analysis is that the ESSQ may not improve MRCS outcomes per se, but rather the ESSQ students are a self-selected group of highly motivated students who were more likely to be successful in the MRCS. The programme, however, was also intended to support the academic development of the surgical trainee at an early stage, and it is noteworthy that 116 students who enrolled in the programme since its launch in 2007 continued to complete a postgraduate degree. This has provided them with a strong academic foundation irrespective of their final career intentions. It may encourage some to pursue an additional period of research training, whereas for others, the academic exposure may replace the traditional approach, which has required surgical trainees in the United Kingdom to pursue 2 or 3 years out of surgical programme training to secure a postgraduate qualification that may make them competitive to progress thought-advanced surgical training. Although, at the present time, there is insufficient follow-up of MSc postgraduates to determine whether their participation will impact on their future surgical training plans or academic development, it is encouraging to note that 26 full research articles have been published directly from content relating to the MSc dissertation project.

nt time, there is insufficient follow-up of MSc postgraduates to determine whether their participation will impact on their future surgical training plans or academic development, it is encouraging to note that 26 full research articles have been published directly from content relating to the MSc dissertation project. There is considerable potential for distance learning programmes to have a major impact on the future delivery of surgical training. The data from this study are relatively immature to determine whether it has contributed to a major change, but it has been demonstrated that the online ESSQ programme supports the academic development of trainees in the early years of surgical training, is well received by students, and is associated with improved success in their professional examination. ACKNOWLEDGMENTS The authors thank the surgical trainees for their insightful feedback on the ESSQ programme, those students who gave permission to use their MRCS examination results in this study, and Kaisey Murphy and members of the Examinations team at the RCSEd for kindly sourcing the relevant MRCS data. They thank the many e-tutors involved, both past and present, for their vital contributions to the online activities and assessments of the ESSQ programme. Reprints: O. James Garden, MD, Clinical Surgery, University of Edinburgh, Royal Infirmary, 51 Little France Crescent, Edinburgh EH16 4SA, Scotland. E-mail o.j.garden@ed.ac.uk. Disclosure: The authors declare no conflicts of interest.

ACKNOWLEDGMENTS The authors thank the surgical trainees for their insightful feedback on the ESSQ programme, those students who gave permission to use their MRCS examination results in this study, and Kaisey Murphy and members of the Examinations team at the RCSEd for kindly sourcing the relevant MRCS data. They thank the many e-tutors involved, both past and present, for their vital contributions to the online activities and assessments of the ESSQ programme. Reprints: O. James Garden, MD, Clinical Surgery, University of Edinburgh, Royal Infirmary, 51 Little France Crescent, Edinburgh EH16 4SA, Scotland. E-mail o.j.garden@ed.ac.uk. Disclosure: The authors declare no conflicts of interest. DISCUSSANTS I. Popescu (Bucharest, Romania): First of all thank you very much for the honor of reviewing this paper. Second, congratulations for the nice work in a very important area; that of education. As a comment, in this paper, the authors have explored the impact of a surgical sciences e-learning program in supporting the academic development of surgical trainees for their professional examination. Thus, it has showed that an e-learning platform is associated with improved outcome in professional examination, master in surgical sciences. Students also had significantly higher MRCS pass rates than nonenrolled trainees; however, it is not clear how the surgical trainees were recruited in the program. Did they voluntarily enroll in the program and did they have to pass a prior examination? One might speculate that surgical trainees who were voluntarily enrolled (with or without a prior examination) could be a priori selected students (the most motivated ones and with the best performances of their generation).

gram. Did they voluntarily enroll in the program and did they have to pass a prior examination? One might speculate that surgical trainees who were voluntarily enrolled (with or without a prior examination) could be a priori selected students (the most motivated ones and with the best performances of their generation). Taking into consideration the aforementioned factors, and including the fact that the questions were similar to the MRCS questions, that may bias the results along with the fact that master in surgical science students have gained added knowledge and expertise within the program. So, in my opinion, it is not quite a surprise that those students have had higher MRCS pass scores. So, finally a question, I would like to ask the authors if there was any significant differences at MRCS pass scores between the surgical trainees that graduated, the highest level with master in surgical science degree, and the other types of enrolled students. Thank you very much.

Taking into consideration the aforementioned factors, and including the fact that the questions were similar to the MRCS questions, that may bias the results along with the fact that master in surgical science students have gained added knowledge and expertise within the program. So, in my opinion, it is not quite a surprise that those students have had higher MRCS pass scores. So, finally a question, I would like to ask the authors if there was any significant differences at MRCS pass scores between the surgical trainees that graduated, the highest level with master in surgical science degree, and the other types of enrolled students. Thank you very much. Response from O.J. Garden (Edinburgh, UK): Thank you. You make a number of important points and the first is in terms of which trainees were allowed to enter the program. Clearly, they must have a medical degree, normally be within 5 years of graduation and be based in a supportive training environment. We do accept your point that the format of part of the assessment of our program is similar to the MCQ component of the MRCS examination. Inevitably, this is going to prepare the student better for that examination. I guess that is part of the aim, but we would also like to think that this also results from their engagement with the program over a prolonged period of time through year 1 and year 2. We have not looked specifically at the performance of the students who have gone on to complete the 3 years of the program. I think the numbers are too small, but I suspect that we will find that these are actually top performing students and I would be surprised if they did not perform well in the MRCS examination.

2. We have not looked specifically at the performance of the students who have gone on to complete the 3 years of the program. I think the numbers are too small, but I suspect that we will find that these are actually top performing students and I would be surprised if they did not perform well in the MRCS examination. DISCUSSANTS R.P. O'connell (Dublin, Ireland): James, this is greatly to be welcomed because any investment in the education of our trainees, such as you have done, is greatly welcomed. But following the last question, it is a bit of a self-fulfilling prophecy that if you invest in the education, and your endpoint is how they do in the membership examination, clearly they are going to do better. My question for you is how do you know they are better clinical scientists as was the objective because it is no longer fashionable to do clinical or laboratory research? What you want to do is to produce clinician scientists and how are you going to measure that outcome?

amination, clearly they are going to do better. My question for you is how do you know they are better clinical scientists as was the objective because it is no longer fashionable to do clinical or laboratory research? What you want to do is to produce clinician scientists and how are you going to measure that outcome? Response from O.J. Garden (Edinburgh, UK): I think it is too early in the program to actually assess that parameter specifically. You will appreciate that it is a contentious issue as to whether or not we should be discouraging some trainees from taking time out to go through the traditional route of academic development, and I know that our own trainees in the southeast of Scotland generally do take time out to undertake a period of full-time study toward an MD or a PhD but not all of them go on to become academic surgeons. I think that there needs to be a way of trying to support their academic development, without taking them out of the clinical training program while providing them with a quality-assured academic qualification. Clearly the students are voting with their feet in saying that that is what we actually want. On the contrary, some of the trainees who have been through this program have actually used this as a stepping-stone to full-time research. Only time will tell as to how the program is going to best fit the needs of the modern surgical trainee.

ts are voting with their feet in saying that that is what we actually want. On the contrary, some of the trainees who have been through this program have actually used this as a stepping-stone to full-time research. Only time will tell as to how the program is going to best fit the needs of the modern surgical trainee. DISCUSSANTS J.V. Lanschot (Rotterdam, The Netherlands): Thank you very much. As I recently learned, the scope of the College in Edinburgh is really global. Some of the residents in The Netherlands have tried to participate in your programs and at the end of the day during the examinations they have to really learn what is the correct answer in the UK to some questions, and what the correct answer would have been in The Netherlands. So, they are in a split, they learn things that they should not learn at least in The Netherlands. My question is what percentage of your program is really global and independent of local knowledge and what percentage is in fact more or less useless to international students?

answer would have been in The Netherlands. So, they are in a split, they learn things that they should not learn at least in The Netherlands. My question is what percentage of your program is really global and independent of local knowledge and what percentage is in fact more or less useless to international students? Response from O.J. Garden (Edinburgh, UK): I am surprised to find out that the surgical evidence base in The Netherlands is different from the evidence base in Great Britain and Ireland, but you do make an important point. This examination, this academic qualification, was very much structured around training in Great Britain and Ireland, and we have been taken quite by surprise that there has been so much interest internationally, but most of the trainees outside of Great Britain and Ireland obviously feel that the content of the program is actually fit for their purpose. It is not delivered in a very didactic way, it is there to support their reflective learning and as we know, from this very meeting, there are many views sometimes on how a particular condition should be managed, when a particular operation should be performed. So, we like to think that what we are trying to do is to develop that academic ethos, to look more critically at the evidence that is there.

e learning and as we know, from this very meeting, there are many views sometimes on how a particular condition should be managed, when a particular operation should be performed. So, we like to think that what we are trying to do is to develop that academic ethos, to look more critically at the evidence that is there. DISCUSSANTS N. Senninger (Munster, Germany): With reference to control mechanisms for your testing, I am convinced that you are going to be successful but you will have to prove it. I would be interested to see what happened, concerning examination performance, to the so-called dropouts or to the ones who were not permitted to the test. What happened to them in comparison to the ones who passed the test successfully? Response from O.J. Garden (Edinburgh, UK): That again is a very important area to try and pursue. The withdrawal rate is lower than many distance learning programs and the predominant reason for students withdrawing is for either financial reasons or a conscious decision to change career such that the programme becomes less relevant to them. Unfortunately, a lot of students who do drop out do not really come back to us when we are looking for information as to why they have left the program and what they have then gone on to do.

After the tragic death of a young woman named Libby Zion in a New York Hospital, resident duty hour (RDH) restrictions were introduced in this state. An investigation ruled that lack of supervision, the resident's lack of familiarity with this woman's complex diagnosis and resident fatigue were contributing factors in the tragic outcome. In July 2003, the Accreditation Council for Graduate Medical Education (ACGME) mandated an 80-hour duty limit on residents, averaged over a 4-week period. In 2011, the ACGME mandated 16-hour duty periods for first year residents whereas the province of Quebec instituted 16-hour periods for all in-house residents in 2012. The stated goals of duty hour limitations were to improve patient safety, resident well-being, and education.1 Surgical disciplines have expressed concern with the contraction of duty hours stating that the impact on training time could have a negative effect on educational and patient outcomes.2 Specifically, the American College of Surgeons, Division of Education, has stated that mastery in surgery requires extensive and immersive experiences.3 A “one-size fits all” approach to RDHs may not be appropriate given the variation in training needs, diversity of practice patterns, and various competencies required among disciplines.4,5

rican College of Surgeons, Division of Education, has stated that mastery in surgery requires extensive and immersive experiences.3 A “one-size fits all” approach to RDHs may not be appropriate given the variation in training needs, diversity of practice patterns, and various competencies required among disciplines.4,5 In March 2012, the Royal College of Physicians and Surgeons of Canada undertook a project to develop a pan-Canadian consensus on issues related to RDH. A national steering committee was struck and this committee further created 6 expert working groups. Recognizing that surgical training is unique in many aspects and that there have been significant obstacles related to the implementation of RDH in surgery, a specific working group entitled Special Considerations for Procedural and Surgical Disciplines was created. The National Steering Committee's full report is published elsewhere (www.residentdutyhours.ca).5 As part of this process, the Procedural/Surgical discipline working group performed a systematic review and meta-analysis to evaluate the association between RDH and clinical and educational outcomes in surgery.

s created. The National Steering Committee's full report is published elsewhere (www.residentdutyhours.ca).5 As part of this process, the Procedural/Surgical discipline working group performed a systematic review and meta-analysis to evaluate the association between RDH and clinical and educational outcomes in surgery. METHODS Study Selection A systematic review (1980–2013) was executed using indexed databases CINAHL, Cochrane Databases, Embase, Medline, and Scopus to identify pertinent articles relating to RDH restrictions with a focus on identifying articles that addressed the 2011 ACGME regulation and Quebec arbitration. Search terms were developed and revised by experts in surgical education. Predetermined eligibility criteria were used to eliminate articles beyond the scope of this review. Articles were included if the research focus was related to the impact of RDH restrictions on surgical resident well-being, education, and/or patient safety. Articles were also included if they addressed heterogeneity across different residency training programs and the varying impacts of RDH on different fields of health care. All abstracts and titles were reviewed by 2 members of the research team for inclusion in review (K.D. and I.K.). To ensure the comprehensive inclusion of high-quality literature on this topic, additional articles identified from hand searching of references and recommendations from content experts were added (Fig. 1). FIGURE 1. Study selection and inclusion criteria.

METHODS Study Selection A systematic review (1980–2013) was executed using indexed databases CINAHL, Cochrane Databases, Embase, Medline, and Scopus to identify pertinent articles relating to RDH restrictions with a focus on identifying articles that addressed the 2011 ACGME regulation and Quebec arbitration. Search terms were developed and revised by experts in surgical education. Predetermined eligibility criteria were used to eliminate articles beyond the scope of this review. Articles were included if the research focus was related to the impact of RDH restrictions on surgical resident well-being, education, and/or patient safety. Articles were also included if they addressed heterogeneity across different residency training programs and the varying impacts of RDH on different fields of health care. All abstracts and titles were reviewed by 2 members of the research team for inclusion in review (K.D. and I.K.). To ensure the comprehensive inclusion of high-quality literature on this topic, additional articles identified from hand searching of references and recommendations from content experts were added (Fig. 1). FIGURE 1. Study selection and inclusion criteria. Data Abstraction and Analysis Quantitative and qualitative data extractions were performed independently by 2 members of the research team for all studies meeting the inclusion criteria. If consensus could not be reached, disputes were resolved by a third member of the research team (N.A.). The quality of the articles was assessed using the GRADE guidelines.6 Each article was rated on a 4-point scale from high to very low, based on the level of confidence in the effect estimate. Attributions for consideration of high quality included multicentre designs, high volume of participants (>100), and control groups and that of low quality included unvalidated surveys and single institutional studies.

ed on a 4-point scale from high to very low, based on the level of confidence in the effect estimate. Attributions for consideration of high quality included multicentre designs, high volume of participants (>100), and control groups and that of low quality included unvalidated surveys and single institutional studies. Extracted data were tabulated and examined for interpretation. Articles identified as moderate and high quality were analyzed separately. Articles graded as low or very low quality were not included in our overall analysis to maintain the integrity of the results. All articles examining 16-hour and night float shifts were analyzed and reported separately to determine the effects of more recent changes. Major findings for patient safety, resident well-being, and education were divided into actual and perceived impact. The term “unclear” was used to categorize studies that reported contradictory results or when the abstractors were unable to draw firm conclusions from the article. Qualitative analysis was completed using the grounded theory approach. Open and axial coding was executed to capture themes from articles relevant to the 3 areas identified as sensitive to RDH restrictions: patient safety, resident well-being, and education. Themes were tabulated and refined. Qualitative themes are reported to add context to the quantitative data. Detailed results of the qualitative review will be reported separately.

hemes from articles relevant to the 3 areas identified as sensitive to RDH restrictions: patient safety, resident well-being, and education. Themes were tabulated and refined. Qualitative themes are reported to add context to the quantitative data. Detailed results of the qualitative review will be reported separately. Statistical Analysis Articles that examined mortality data were combined in a random-effects meta-analysis to evaluate the impact of RDH on patient outcomes. The analysis was conducted in the statistical program R using the metafor package.7,8 An odds ratio greater than 1 means a higher risk of death for Pre-RDH, relative to Post-RDH and odds ratio less than 1 means a lower risk of death for Pre-RDH relative to Post-RDH. A random-effects model was fit using the restricted maximum likelihood estimator. RESULTS A total of 709 articles were identified from indexed sources (CINAHL: 50, Cochrane Databases: 4, Embase: 203, Medline: 301, Scopus: 151). After the review of abstracts, full articles, suggested papers from content experts, and hand searching of references 135 articles were identified as eligible for data analysis (Fig. 1).

tal of 709 articles were identified from indexed sources (CINAHL: 50, Cochrane Databases: 4, Embase: 203, Medline: 301, Scopus: 151). After the review of abstracts, full articles, suggested papers from content experts, and hand searching of references 135 articles were identified as eligible for data analysis (Fig. 1). Study Characteristics Ninety-five percent (128) of the studies were conducted in the United States or Canada, with 53% of these being multicentre (Table 1). General surgery was the most commonly studied discipline, comprising half of all studies. The primary study designs were surveys (61, 45%), interventional studies, with the intervention being the implementation of RDH restrictions (52, 39%) and observational studies (8, 6%). Only 1 article used a randomized controlled trial to examine the impact of duty hour restrictions (1%). Overall study objectives were to evaluate the impact of duty hour restrictions on resident education (101, 75%), patient safety (79, 59%), and resident wellness (48, 36%). Nine articles (7%) specifically compared the effect of RDH on training in various specialties. Fifty-eight percent of articles were graded as low or very low quality (N = 78) and 42% of studies were graded as moderate or high quality (N = 57). TABLE 1. Characteristics of Articles (N = 135)

Study Characteristics Ninety-five percent (128) of the studies were conducted in the United States or Canada, with 53% of these being multicentre (Table 1). General surgery was the most commonly studied discipline, comprising half of all studies. The primary study designs were surveys (61, 45%), interventional studies, with the intervention being the implementation of RDH restrictions (52, 39%) and observational studies (8, 6%). Only 1 article used a randomized controlled trial to examine the impact of duty hour restrictions (1%). Overall study objectives were to evaluate the impact of duty hour restrictions on resident education (101, 75%), patient safety (79, 59%), and resident wellness (48, 36%). Nine articles (7%) specifically compared the effect of RDH on training in various specialties. Fifty-eight percent of articles were graded as low or very low quality (N = 78) and 42% of studies were graded as moderate or high quality (N = 57). TABLE 1. Characteristics of Articles (N = 135) Country Study design United States 121 (90)  Survey 61 (46) Canada   7 (5)  Interventional (≥2 groups) 52 (39) Europe 6 (4)  Observational 8 (6) New Zealand 1 (1)  Systematic review 6 (4) Type of center  Other 5 (4) Academic 86 (64)  Program evaluation 2 (1) Community 20 (15)  Randomized controlled trial 1 (1) Multicentre 71 (53) Surgical subspecialty Study objectives General surgery 67 (50)  Education 101 (75) Orthopedic 15 (11)  Patient safety 79 (59) Gynecology 13 (10)  Wellness 48 (36) Multi-specialty (including surgery) 12 (9)  Heterogeneity 9 (7) All surgical specialties 12 (9) Study quality* Trauma 10 (7)  High 21 (15) Other† 10 (7)  Moderate 36 (27) ENT 8 (6)  Low 51 (38) Pediatrics 8 (6)  Very low 27 (20) Cardiac 8 (6) Type of surgery Neurosurgery 6 (4)  Emergency 95 (70) Vascular 5 (4)  Elective 88 (65) Thoracic 5 (4)  Not described 38 (28) Urology 3 (2) Plastic 3 (2) Not described 12 (9) All the values indicate n (%).

igh 21 (15) Other† 10 (7)  Moderate 36 (27) ENT 8 (6)  Low 51 (38) Pediatrics 8 (6)  Very low 27 (20) Cardiac 8 (6) Type of surgery Neurosurgery 6 (4)  Emergency 95 (70) Vascular 5 (4)  Elective 88 (65) Thoracic 5 (4)  Not described 38 (28) Urology 3 (2) Plastic 3 (2) Not described 12 (9) All the values indicate n (%). *Graded moderate-high quality as per Balshem et al.6 †Other categories include nonsurgical disciplines reported in combination with surgical subspecialties and infrequently reported surgical specialties (ophthalmology, transplant, and critical care). Major findings of articles with respect to the impact of duty hour restrictions on patient safety, resident well-being, and education can be listed as follows (Table 2). TABLE 2. Summary of Effect of RDH Restrictions on Patient Safety and Educational Outcomes for Studies Graded as Moderate-High Quality, and for Studies Focusing on 16-Hour Duty Maximum and/or Night Float Systems

Major findings of articles with respect to the impact of duty hour restrictions on patient safety, resident well-being, and education can be listed as follows (Table 2). TABLE 2. Summary of Effect of RDH Restrictions on Patient Safety and Educational Outcomes for Studies Graded as Moderate-High Quality, and for Studies Focusing on 16-Hour Duty Maximum and/or Night Float Systems Improved Worsened No Change Unclear† Total Effect of RDH Restrictions* N % N % N % N % N Section 1–-Moderate-high quality articles (N = 57)‡ Patient safety 5 19 8 30 13 48 1 4 27 Wellness 5 71 0 0 2 29 0 0 7 Education 1 4 13 48 11 41 2 7 27 Perception of wellness (survey) 4 57 0 0 3 43 0 0 7 Perception of education (survey) 1 9 4 36 4 36 2 18 11 Perception of patient safety (survey) 0 0 6 67 2 22 1 11 9 Section 2—16-h duty maximums (N = 17) Patient safety 0 0 0 0 1 50 1 50 2 Wellness 0 0 0 0 2 100 0 0 2 Education 0 0 2 100 0 0 0 0 2 Perception of wellness (survey) 3 38 1 13 1 13 3 38 8 Perception of education (survey) 0 0 12 86 2 14 0 0 14 Perception of patient safety (survey) 0 0 11 79 2 14 1 7 14 Section 3–-Night float systems (N = 8) Patient safety 0 0 0 0 1 50 1 50 2 Wellness 1 100 0 0 0 0 0 0 1 Education 0 0 1 50 0 0 1 50 2 Perception of wellness (survey) 3 100 0 0 0 0 0 0 3 Perception of education (survey) 1 25 1 25 0 0 2 50 4 Perception of patient safety (survey) 1 33 2 67 0 0 0 0 3 Section 4–-Both 16-h duty maximums and night float systems (N = 24) Patient safety 0 0 0 0 2 67 1 33 3 Wellness 1 33 0 0 2 67 0 0 3 Education 0 0 2 67 0 0 1 33 3 Perception of wellness (survey) 6 55 1 9 1 9 3 27 11 Perception of education (survey) 1 6 12 71 2 12 2 12 17 Perception of patient safety (survey) 1 6 12 75 2 13 1 6 16 *RDH restrictions including 2003, 2008 ACGME guidelines and European working time directive guidelines. Four articles in this review assessed the impact on nonsurgical residents and surgical residents, 3 were included in the high quality review,20,34,43 3 were included in the review of the 16-hour max shift,34,70,72 and 1 was included in the review of the night float system.43

European working time directive guidelines. Four articles in this review assessed the impact on nonsurgical residents and surgical residents, 3 were included in the high quality review,20,34,43 3 were included in the review of the 16-hour max shift,34,70,72 and 1 was included in the review of the night float system.43 †Contradictory results or unable to draw firm conclusions from findings. ‡Graded moderate-high quality as per Balshem et al. (2011).6 Section 1: Effect of RDH Restrictions Including 2003, 2011 ACGME Restrictions and the European Working Time Directive, From 57 Moderate to High Quality Studies Patient Safety. Thirty-five (61%) articles, graded moderate to high quality, examined patient safety; 8 (30%) of these reported worsened patient safety outcomes (morbidity and mortality) as a result of duty hour restrictions.9–16 Five (19%) reported an improvement17–21 and 13 (48%) reported no change.22–34 Of the 9 articles evaluating perceptions of RDH impact on patient safety, 6 (67%) reported a decrease in quality of care,35–40 and 2 (22%) reported no change.41,42 Thus, the current impact on patient safety is inconclusive, but perceptions of health care professionals suggest negative outcomes.

reported no change.22–34 Of the 9 articles evaluating perceptions of RDH impact on patient safety, 6 (67%) reported a decrease in quality of care,35–40 and 2 (22%) reported no change.41,42 Thus, the current impact on patient safety is inconclusive, but perceptions of health care professionals suggest negative outcomes. Resident Well-being. Of the 57 articles considered moderate to high quality, 9 (16%) examined resident wellness. Five (71%) reported that duty hour restrictions had improved resident wellness20,24,43–45 and 2 (29%) reported no change,34,37 as measured by validated tools such as Maslach Burnout Inventory. Of the 7 articles evaluating perception of wellness, 4 (57%) reported improvement after RDH24,42,44 and 3 (43%) reported no change.37,39,45 Overall, the majority of articles reported improved resident wellness.

,43–45 and 2 (29%) reported no change,34,37 as measured by validated tools such as Maslach Burnout Inventory. Of the 7 articles evaluating perception of wellness, 4 (57%) reported improvement after RDH24,42,44 and 3 (43%) reported no change.37,39,45 Overall, the majority of articles reported improved resident wellness. Resident Education. Thirty-three (58%) articles graded as moderate to high quality examined resident education. Thirteen (48%) of these reported that educational outcomes were worsened after the implementation of RDH38,40,46–56 and 11 (41%) reported no change,22,24,26,32,45,57–62 using measures such as examination scores including American Board of Surgery certification data. Only 1 (4%) study showed an improvement in educational outcomes; the balance of studies showed worsened or no significant effects of RDH restrictions.44 Of the 11 articles evaluating self-reported perceptions of RDH impact on education, 4 (36%) reported a decrease in quality of the educational environment after RDH36,38–40 and 4 (36%) reported no change.35,37,42,61 Only 1 article (9%) reported a perceived improvement in education.43 Therefore, on the basis of objective tools of measurement and perceived impacts, findings suggest resident education has either worsened or remained unchanged after duty hour restrictions. Section 2: Effect of the 16-Hour Duty Maximums on Patient Safety, Resident Well-being and Educational Outcomes From 17 Articles Seventeen studies evaluated the impact of 16-hour duty maximums; 11 studies were low quality, 6 were moderate to high quality (Table 2).

Resident Education. Thirty-three (58%) articles graded as moderate to high quality examined resident education. Thirteen (48%) of these reported that educational outcomes were worsened after the implementation of RDH38,40,46–56 and 11 (41%) reported no change,22,24,26,32,45,57–62 using measures such as examination scores including American Board of Surgery certification data. Only 1 (4%) study showed an improvement in educational outcomes; the balance of studies showed worsened or no significant effects of RDH restrictions.44 Of the 11 articles evaluating self-reported perceptions of RDH impact on education, 4 (36%) reported a decrease in quality of the educational environment after RDH36,38–40 and 4 (36%) reported no change.35,37,42,61 Only 1 article (9%) reported a perceived improvement in education.43 Therefore, on the basis of objective tools of measurement and perceived impacts, findings suggest resident education has either worsened or remained unchanged after duty hour restrictions. Section 2: Effect of the 16-Hour Duty Maximums on Patient Safety, Resident Well-being and Educational Outcomes From 17 Articles Seventeen studies evaluated the impact of 16-hour duty maximums; 11 studies were low quality, 6 were moderate to high quality (Table 2). Patient Safety. Fifteen articles examined the impact of the 16-hour duty maximum restrictions on patient safety. Two articles objectively analyzed the impact (eg, examined morbidity and mortality rates), 1 study had unclear results38 and the other reported no change.34 Of the 14 articles that evaluated the residents' perceptions of the impact of 16-hour duty maximum on patient safety, 11 (79%) reported that patient safety has declined,4,36–38,63–69 2 (14%) suggested there has been no effect,70,71 and 1 (7%) was inconclusive.72 Thus, the objective impact on patient safety is inconclusive, but the perceived impact suggests patient safety has declined. Decreased continuity of care and increased number of handovers were themes identified from the qualitative review as causes for a decline in patient outcomes after duty hour restrictions.

sive.72 Thus, the objective impact on patient safety is inconclusive, but the perceived impact suggests patient safety has declined. Decreased continuity of care and increased number of handovers were themes identified from the qualitative review as causes for a decline in patient outcomes after duty hour restrictions. Resident Wellness, Fatigue, and Burnout. Nine of the 17 articles examined the impact of 16-hour duty hour restrictions on resident wellness; 2 articles using validated measures (eg, Maslach Burnout Inventory) concluded there has been no significant change. Eight articles examined the perceived impact; 3 (38%) reported improvement,64,70,72 1 (13%) suggested wellness was worse,67 1 (13%) reported no significant change,37 and 3 (38%) had unclear results.4,65,71 Validated tools suggest no significant change after RDH restrictions; findings from the perceived impact are unclear. Thematic review of the data suggests resident training, professional satisfaction, preparedness for more senior roles, and the wellness of more senior trainees has been compromised by the practice of 16-hour duty maximums.71

t no significant change after RDH restrictions; findings from the perceived impact are unclear. Thematic review of the data suggests resident training, professional satisfaction, preparedness for more senior roles, and the wellness of more senior trainees has been compromised by the practice of 16-hour duty maximums.71 Resident Education. Fifteen articles evaluated impact of the 16-hour duty maximums on resident education; 2 objectively assessed the impact and concluded education was worse (eg, examination data and/or clinical performance).38,55 Fourteen examined the perceived impact; 12 (86%) articles concluded that education is worse4,36,38,64–72 and 2 (14%) reported no significant change.22,35,37 Overall, both the objective and perceived impact findings support a trend toward worsened outcomes. Thematic data analysis of the literature suggests 16-hour duty maximums result in poorer integration into teams and mentoring of residents, increased patient handover, decreased operative exposure, and increased self-reported medical errors.

ective and perceived impact findings support a trend toward worsened outcomes. Thematic data analysis of the literature suggests 16-hour duty maximums result in poorer integration into teams and mentoring of residents, increased patient handover, decreased operative exposure, and increased self-reported medical errors. Section 3: Effect of Night Float Systems From 8 Studies Our analysis of the 8 studies examining night float systems showed that there was an improvement in resident wellness reported in 1 study,43 there were 2 studies that reported adverse or unclear effects on resident education,38,43 and 2 studies reported negative impacts on patient safety (Table 2).29,38 No studies of night float systems showed improved patient or resident educational outcomes using objective criteria. Survey data suggested improved resident wellness (N = 3 studies).43,73,74 Two studies reported a perception of worsening patient outcomes38,73 with an inconclusive effect on resident education.43,73

9,38 No studies of night float systems showed improved patient or resident educational outcomes using objective criteria. Survey data suggested improved resident wellness (N = 3 studies).43,73,74 Two studies reported a perception of worsening patient outcomes38,73 with an inconclusive effect on resident education.43,73 Section 4: Effect of 16-Hour Duty Maximums and Night Float Systems From 24 Studies When considering the 24 articles examining both 16-hour duty maximums and night float systems, we found that 3 studies examined the effect of these call models on morbidity and mortality (Table 2).29,34,38 No studies showed an improvement or a worsening effect on patient safety as a result of these changes to RDH. Concerning resident wellness, 1 of 3 studies reported an improvement,43 whereas 2 studies reported no change in resident well-being.34,37 With respect to resident education, there were no studies showing improvement. Two studies showed worsened educational outcomes using examination data, access to operating room (OR) times, and clinical performance;38,55 in 1 study, the overall impact could not be determined.43 In more than half of the studies (6, 55%) using self-reported survey data, residents believed that their wellness had improved after 16-hour shift restrictions and/or the implementation of night float systems.43,64,70,72–74 Seventeen surveys queried self-reported educational outcomes and 12 (71%) surveys showed that educational outcomes were believed to have worsened;4,36,38,64–72 only 1 (6%) survey reported improved educational outcomes.73 Sixteen survey studies examined the effects of 16-hour or night float systems on perceptions of patient safety; 12 (75%) reported a perception of worsened patient outcomes,4,35–38,64–69,73 whereas 1 (6%) survey reported improved perceptions of patient safety.74

ly 1 (6%) survey reported improved educational outcomes.73 Sixteen survey studies examined the effects of 16-hour or night float systems on perceptions of patient safety; 12 (75%) reported a perception of worsened patient outcomes,4,35–38,64–69,73 whereas 1 (6%) survey reported improved perceptions of patient safety.74 Figure 2 summarizes the findings from the 17 papers related to 16-hour duty maximums; 13 studies were surveys. The results of studies related to resident wellness and fatigue are heterogeneous with a few studies in each category. Six studies examined time in the OR and overall reported that there was decreased time in the OR (N = 5) as a result of RDH restrictions.4,35,37,55,68 Fifteen studies examined the effect on patient safety and found a strong signal toward worsened patient outcomes (N = 10),4,35–37,64–69 with no study reporting improved patient outcomes. Fifteen studies examined education and training; 12 of these showed worsened outcomes4,36,38,55,65–68,70–72 and 3 studies showed no significant change.35,37,69 No studies reported improved educational outcomes. FIGURE 2. Impact of 16-hour duty maximums in surgery (N = 17 studies).

Figure 2 summarizes the findings from the 17 papers related to 16-hour duty maximums; 13 studies were surveys. The results of studies related to resident wellness and fatigue are heterogeneous with a few studies in each category. Six studies examined time in the OR and overall reported that there was decreased time in the OR (N = 5) as a result of RDH restrictions.4,35,37,55,68 Fifteen studies examined the effect on patient safety and found a strong signal toward worsened patient outcomes (N = 10),4,35–37,64–69 with no study reporting improved patient outcomes. Fifteen studies examined education and training; 12 of these showed worsened outcomes4,36,38,55,65–68,70–72 and 3 studies showed no significant change.35,37,69 No studies reported improved educational outcomes. FIGURE 2. Impact of 16-hour duty maximums in surgery (N = 17 studies). Section 5: Effect of RDH Restrictions on Patient Mortality and Morbidity Mortality. A Forest plot summarizing 13 studies examining the effect of RDH on patient mortality is presented in Figure 3. Two population-based series demonstrate an increase in patient mortality.18,19 Overall, the Random Effects model suggests a trend toward increased mortality in surgical patients in the epoch of time after the implementation of RDH restrictions, compared with the epoch of time before the implementation of RDH restrictions. However, this effect did not reach statistical significance. FIGURE 3. Forest plot describing the effect of RDH restrictions on patient mortality. TABLE 3. Effect of RDH Restrictions on Postoperative Morbidity

Section 5: Effect of RDH Restrictions on Patient Mortality and Morbidity Mortality. A Forest plot summarizing 13 studies examining the effect of RDH on patient mortality is presented in Figure 3. Two population-based series demonstrate an increase in patient mortality.18,19 Overall, the Random Effects model suggests a trend toward increased mortality in surgical patients in the epoch of time after the implementation of RDH restrictions, compared with the epoch of time before the implementation of RDH restrictions. However, this effect did not reach statistical significance. FIGURE 3. Forest plot describing the effect of RDH restrictions on patient mortality. TABLE 3. Effect of RDH Restrictions on Postoperative Morbidity Reference Quality Score Discipline Participants Morbidity Measure Outcome Values* Impact of RDH on Morbidity† Interventional study design Pre-RDH Post-RDH Browne et al14 High Orthopedic surgery 48,430 Perioperative complications NA NA Increased De Virgilio et al22 Moderate Trauma surgery 11,518 Overall morbidity rate 38 40 No change Dumont et al10 Moderate Neurosurgery 12,957 Morbidity rates per 1000 patients 70 89 Increased Avoidable or possibly preventable morbidity rates per 100 patients 56 76 Increased Gopaldas et al21 Moderate Cardiothoracic surgery 1562 Ventilator time >48 h 60 93 Increased Reoperation for bleeding 5 24 Increased Gopaldas et al13 High Cardiac surgery 614,177 In-hospital complications 78446 50390 Increased Hoh et al11 High Neurosurgery 222,610 Morbidity rates 483 832 Increased Kaderli et al12 Moderate Surgery 2686 Postoperative surgical complications 32 72 Increased Intraoperative complications 6 8 No change Morrison et al15 High Trauma surgery 492,173 Mechanical ventilation 1.6 d 1.3 d Decreased Length of ICU stay 2.2 2.1 Decreased Length of hospital stay 5.8 5.8 No change Privette et al18 Moderate Trauma surgery; Vascular surgery; General surgery 14,610 Total complications 1688 1852 No change Major complications 333 212 Decreased Moderate complications 690 768 No change Minor complications 663 869 Increased Salim et al9 Moderate Trauma surgery 16,854 Total number of complications 504 576 No change Number of preventable complications 80 101 No change Schenarts et al29 Moderate Trauma surgery 2826 Respiratory, cardiac, intra-abdominal, and infectious complications 391 401 No change Length of ICU stay 5.4 d 5.4 d No change Mechanical ventilation 5.8 d 5.6 d No change Length of hospital stay 6.8 d 6.4 d No change Yaghoubian et al19 Moderate General surgery 2470 Overall morbidity rate 65 24 Decreased Systematic reviews Baldwin et al17 Moderate Surgery and Medicine NA 13 studies assessed morbidity rates: deep surgical infections, disseminated intravascular coagulation, intra-abdominal abscesses, mediastinitis, medication errors, complications attributed to care providers, electrolyte abnormalities, drug-drug interactions, hematomas, transfusion

al17 Moderate Surgery and Medicine NA 13 studies assessed morbidity rates: deep surgical infections, disseminated intravascular coagulation, intra-abdominal abscesses, mediastinitis, medication errors, complications attributed to care providers, electrolyte abnormalities, drug-drug interactions, hematomas, transfusion s, renal complications, reoperations for bleeding, thromboembolic complications, septicemia, test delays, and systems and patient-related complications. NA NA 5 studies demonstrated decreases in individual complications; 7 studies demonstrated increases in individual complications Fletcher et al20 High Surgery‡ NA 11 articles assessed morbidity rates; overall complication rates and readmissions to the surgical intensive care unit NA NA 2 studies showed a decrease in overall complications, 2 showed an increase in complication rates, 3 revealed no significant change and the outcomes were not available in 4 studies. Jamal et al25 High Surgery 64,346 Morbidity rates 33317 31029 No change *If comparison between teaching and nonteaching hospitals, reported values are from teaching hospitals. †Studies reporting an increase or decrease in morbidity showed a significant difference in their findings; studies with no change did not find a statistically significant difference before and after the implementation of RDH restrictions. ‡Data from surgical studies only. NA indicates not available.

NA NA 5 studies demonstrated decreases in individual complications; 7 studies demonstrated increases in individual complications Fletcher et al20 High Surgery‡ NA 11 articles assessed morbidity rates; overall complication rates and readmissions to the surgical intensive care unit NA NA 2 studies showed a decrease in overall complications, 2 showed an increase in complication rates, 3 revealed no significant change and the outcomes were not available in 4 studies. Jamal et al25 High Surgery 64,346 Morbidity rates 33317 31029 No change *If comparison between teaching and nonteaching hospitals, reported values are from teaching hospitals. †Studies reporting an increase or decrease in morbidity showed a significant difference in their findings; studies with no change did not find a statistically significant difference before and after the implementation of RDH restrictions. ‡Data from surgical studies only. NA indicates not available. Morbidity. Fifteen studies examined the effect of RDH restrictions on postoperative morbidity and are summarized in Table 3. Because of the heterogeneity of the manner in which complications reported, we were unable to conduct a meta-analysis related to these outcomes. Overall, there were 15 studies from which we could extract data; 6 studies reported increased patient-level morbidity as a result of RDH.10–14,21 The systematic review conducted by Jamal et al25 reported no overall change, but their results suggested an increase in complications in patients undergoing emergency surgery. Two systematic reviews had heterogeneous results and did not report a conclusive impact of RDH restrictions on surgical patient morbidity.17,20 Two studies reported decreased complication rates,15,19 and 3 studies reported no significant change in patient complications as a result of RDH restrictions.5,22,29 Privette et al showed decreased rates for major complications after RDH restrictions but increased rates for minor complications, and no significant change for total and moderate complications after RDH restrictions.18 More recent studies and those studies examining acutely ill patients reported increased complication rates post duty hour regulations.10–13,21

major complications after RDH restrictions but increased rates for minor complications, and no significant change for total and moderate complications after RDH restrictions.18 More recent studies and those studies examining acutely ill patients reported increased complication rates post duty hour regulations.10–13,21 DISCUSSION We have completed a comprehensive review of the relevant literature related to RDH restrictions in surgery. This study has several limitations. These limitations are related to the heterogeneous nature of the studies, the various RDH restrictions imposed over varying time intervals, the clinical contexts in which they were conducted, and that for the vast majority of studies, randomization was not possible owing to the regulated nature of duty hours in most jurisdictions. In addition, although there are many areas of debate with respect to the impact of duty hour restrictions, this article focused on patient safety, resident wellness, and resident education. Unfortunately, other topics such as professionalism, impact on surgical faculty and costs associated with mitigating the impact of RDH restrictions was deemed outside the scope of this review. The large number of articles examining perceptions of RDH impact using unvalidated surveys reduces the validity of the findings; to mitigate this, the perceived impacts were reported separately. Most of these papers were categorized as low quality and therefore not included in our final results. Finally, it proved impossible to conduct a meaningful meta-analysis of the literature related to surgical complications because of the manner in which these data were collected and reported across studies. The authors instead chose to report their findings in tabular format.

therefore not included in our final results. Finally, it proved impossible to conduct a meaningful meta-analysis of the literature related to surgical complications because of the manner in which these data were collected and reported across studies. The authors instead chose to report their findings in tabular format. Patient Safety One of the stated goals of RDH restrictions was to protect patients from the deleterious effects of physician fatigue. Although evidence demonstrating deficits in attention and hand-eye coordination after sleep deprivation suggest that physician fatigue could compromise patient safety, particularly for the surgical population, this contention has not been demonstrated to date.16,22,25–27,33,75–79 Careful review of the literature over the last decade fails to demonstrate improved patient safety after the implementation of RDH restrictions. A recently published, high-level meta-analysis surveying the surgical literatures shows convincingly that there has been no improvement in patient safety as a result of RDH restrictions.25 Our study confirmed that overall there was no reduction in mortality in the RDH epoch (Fig. 2). More recently, there is increasingly strong evidence that RDH restrictions could be associated with harm to patients.11,75 Recent, larger studies demonstrated increased morbidity and mortality, specifically in high-acuity patient populations as a result of restrictions in RDH.10–12 Specifically, studies in neurosurgery, cardiac surgery, and critical care show that patient-level complications have increased after the implementation of restricted RDH.10,11,13 The largest study suggesting that RDH restrictions have increased patient morbidity was performed by Poulose et al.16 An interrupted time series analysis was performed between 1995 and 2001. Approximately 2.6 million patient discharges from New York area hospitals were reviewed annually over that time period. The authors demonstrated increased incidence of thrombotic/embolic complications and accidental lacerations and punctures after the institution of RDH restrictions. In 2012, Hoh et al11 analyzed the Nationwide Inpatient Sample database of more than 107,000 neurotrauma patients. They compared teaching hospital versus nonteaching hospitals for a 3-year period before and after the 2003 ACGME Work Duty Hours restrictions.

ations and punctures after the institution of RDH restrictions. In 2012, Hoh et al11 analyzed the Nationwide Inpatient Sample database of more than 107,000 neurotrauma patients. They compared teaching hospital versus nonteaching hospitals for a 3-year period before and after the 2003 ACGME Work Duty Hours restrictions. The complication rates did not change in nonteaching hospitals (OR = 0.98; 95% CI: 0.787–1.22; P = 0.86) but did change significantly in teaching hospitals (OR = 1.23; 95% CI: 1.00–1.50; P < 0.05). There were no differences in mortality rates between the hospitals across these time points. The increased complications were attributed to the increased handovers. The notion that increasing the frequency of handovers has negative impacts on patient safety has been substantiated by a number of other studies.63,76,80 A recent study published by Dumont et al in the Journal of Neurosurgery also suggested an increase in complications. The authors studied a prospectively collected divisional Morbidity and Mortality database. They showed that the morbidity rate increased from 70 to 89 per 1000 patients treated after institution of work-hour restrictions (P = 0.001). The morbidities that were deemed avoidable or possibly preventable increased from 56 to 66 per 1000 patients treated (P = 0.017).10 These same authors queried the Nationwide Inpatient Sample for all patients admitted for elective craniotomy for meningioma from 1998 to 2008, excluding 2003. The complication rate was compared for 5-year epochs at teaching and nonteaching hospitals before (1998–2002) and after (2004–2008) the ACGME work hour restrictions. They identified 21,177 patients who met the inclusion criteria. They found a statistically significant increase in complication rates from 14% to 16% in teaching hospitals in contrast to a stable 15% complication rate in nonteaching hospitals. This effect remained significant in a multivariate analysis including age and existing comorbidities. Aligned with previous research related to the 2003 ACGME regulations, there is no demonstrable improvement in patient safety after the 2011 ACGME 16-hour duty limitations.

5% complication rate in nonteaching hospitals. This effect remained significant in a multivariate analysis including age and existing comorbidities. Aligned with previous research related to the 2003 ACGME regulations, there is no demonstrable improvement in patient safety after the 2011 ACGME 16-hour duty limitations. More recent evidence studying the effect of 2011 ACGME regulations suggests a decrease in patient safety related to increased numbers of handovers and work compression.34 Similar results of increased complications as a result of RDH restrictions have been documented in a large series of intensive care unit (ICU) patients.81 Surgical diseases, complication, and outcomes may differ qualitatively from other fields of medicine, as surgical practice is tied to large inpatient units with patients whose clinical course can evolve significantly over hours. The observed worsening of outcomes in surgery as a result of RDH restrictions may be related to an increased frequency of handovers, which cause fragmented care and loss of crucial information. Frequent shift changes, as experienced with some restrictive RDH stipulations, may impede residents' ability to observe the natural course of disease and recognize deviations from an expected trajectory and complications of surgery. It is imperative that surgeons are trained to recognize the early signs of postoperative problems and institute appropriate interventions. The development of surgical acumen and the management of complications of surgery require repeated exposures so that subtle variations in clinical presentations can be mastered.

ry. It is imperative that surgeons are trained to recognize the early signs of postoperative problems and institute appropriate interventions. The development of surgical acumen and the management of complications of surgery require repeated exposures so that subtle variations in clinical presentations can be mastered. Resident Wellness, Fatigue, and Burnout Review of the literature pertaining to the change in RDH restrictions in surgery shows an improvement in resident well-being, fatigue, and burnout after the implementation of the 80-hour workweek and coincident with the 2003 ACGME regulations. However, the anticipated improvements in resident wellness, physical or occupational health, have not been consistently realized from subsequent regulations, such as the 2011 ACGME stipulations of 16-hour in-house duty maximums for first-year residents. A longitudinal cohort study undertaken by Sen et al34 showed that the decrease of working hours under the 2011 ACGME standards have resulted in an “unanticipated increase in self-reported medical errors” and has not shown an improvement in well-being or depressive symptoms, nor has it prompted an increase in hours of sleep. Drolet et al in a large national survey of residents from all specialties showed that 48% of respondents disapprove of 16-hour call maximums and only 22% of respondents supported the move to 16-hour duty in-house call maximums. The restrictions were perceived as detrimental to resident quality of life, education, and engagement with their supervisor. The residents' survey responses suggest an improved quality of life for interns; however, quality of life for senior residents seems to suffer in the new paradigm. Interns also reported a decreased sense of readiness for more senior roles, raising a concern that 2011 duty hour reforms could result in prolongation of an already lengthy training.71,72 Antiel and colleagues conducted a longitudinal survey of US PGY1 (post graduate year) surgery residents training under the 16-hour duty maximum regulations.

decreased sense of readiness for more senior roles, raising a concern that 2011 duty hour reforms could result in prolongation of an already lengthy training.71,72 Antiel and colleagues conducted a longitudinal survey of US PGY1 (post graduate year) surgery residents training under the 16-hour duty maximum regulations. Participants reported a significant negative effect on quality of life in their first year of training, a lack of continuity of care, exposure to supervisors and operative cases.37 A prospective longitudinal cohort multicentre study of 2323 PGY1 trainees (58% response rate) showed that although there was a small decrease in duty hours worked per week, there was no significant change in hours slept, depressive symptoms, or well-being (measured using a validated tool).34 A smaller, single-center study showed that surgical residents experienced a paradoxical increased sense of fatigue and felt that their education was compromised as a result of 16-hour duty maximums. In addition to residents reporting feeling more fatigued, they also reported that they could not master surgical skills and perceived a decline in the quality of patient care they provided. Seventy percent of residents reported feeling disconnected from their service and that their mentoring relationship with faculty was compromised. Sixteen hour in house call maximums resulted in increased rates of burnout, isolation, as well as reduced morale among surgical trainees.67

ity of patient care they provided. Seventy percent of residents reported feeling disconnected from their service and that their mentoring relationship with faculty was compromised. Sixteen hour in house call maximums resulted in increased rates of burnout, isolation, as well as reduced morale among surgical trainees.67 Night, evening, rotating, and irregular shifts may also be associated with an elevated risk of workplace injuries and elevated rates of burnout. Shift work in general is associated with increased preterm labor, depression, weight gain, increased rates of injury and burnout and has been classified as a class I carcinogen by the World Health Organization.82 Studies show increased rates of needle stick injury among surgical residents training under traditional 24-hour call models.83 However, the incidence of needle stick injuries is correlated with the level of the learner more so than with sleep deprivation.84 These findings suggest that implementation of early education could decrease this specific occupational risk. Continuous wakefulness for prolonged periods (24 hours) should be avoided, especially if this pattern is frequently repeated. The period of duty for residents may not necessarily have to equate with continuous wakefulness. Rest periods and protected sleep time that respects circadian physiology could improve resident fatigue and professional satisfaction, whereas preserving their access to educational opportunities and preserving patient safety. Night float systems alternate at weekly intervals as per the ACGME guidelines, causing not only a disruption in one's internal balance but creating a cycle where a resident may not have sufficient time to recover from the disruption before being exposed again to a 12-hour shift. Residents who adhere to traditional 24-hours call schedules may be more likely overall to experience relatively normal sleep patterns with deep sleep occurring at night. Thus far, it seems the recent trend toward night float and shift work has not yielded the desired result of more rested and professionally satisfied surgical residents.

re to traditional 24-hours call schedules may be more likely overall to experience relatively normal sleep patterns with deep sleep occurring at night. Thus far, it seems the recent trend toward night float and shift work has not yielded the desired result of more rested and professionally satisfied surgical residents. Education and Training Overall, of the 15 studies examining the effect of 16-hour duty maximums, none showed improvements in educational outcomes after the implementation of RDH restrictions. The majority of studies showed worsening of outcomes, whereas a few studies showed no change. Studies related to RDH restriction in education showed that overall the total numbers of operations performed during residency have decreased slightly. However, there has been a proportionately greater decrease in the volume of emergency surgical procedures performed. Furthermore, the residents' role in the OR has changed over the last decade, with a significant decline in the numbers of cases reported as first assistant and as teaching assistant.51 This trend is worrisome, as first assistant cases represent an opportunity for trainees to learn surgical anatomy, steps of the procedure and receive mentorship in the OR, and the teaching assistant role allows residents to develop and demonstrate surgical mastery.51 These findings are consistent with recent studies related to resident time in the OR after the implementation of the 16-hour duty maximums, with 5 of 6 studies demonstrating decreased time in the OR; no studies demonstrated a neutral effect or increased time in the OR. Most recently, in a study of 249 surgical residents, Schwartz et al55 showed a 25% decrease in PGY1-level operative cases subsequent to the implementation of 16-hour duty maximums, compared to the 4 previous years.

nstrating decreased time in the OR; no studies demonstrated a neutral effect or increased time in the OR. Most recently, in a study of 249 surgical residents, Schwartz et al55 showed a 25% decrease in PGY1-level operative cases subsequent to the implementation of 16-hour duty maximums, compared to the 4 previous years. The mastery of procedural components of surgery requires time on task in a mentored environment and with graduated responsibility. Specifically, achieving mastery of complex technical tasks requires deliberate practice. Some authors suggest that 10,000 hours of deliberate practice may be required.85,86 Teaching of technical expertise and surgical judgement requires that mentors spend longitudinal time with trainees to appreciate their strengths and weaknesses. Coaching in the OR specifically requires that mentors observe progress, provide feedback, and then look for progress at the next opportunity. In teaching programs, surgical faculty voice a growing concern that they have insufficient exposure to individual residents to identify deficiencies, address weaknesses, or provide meaningful evaluation. This observation is especially true for night-float rotations where exposure to teaching faculty can be extremely variable, random, and often brief.87 This phenomenon may delay the maturation of surgical competence in the OR and result in prolongation of training. This unintended consequences of shift work and night float systems was described in a nationwide survey of US PGY1 residents as they are feeling less prepared to assume the PGY2 role.71

, and often brief.87 This phenomenon may delay the maturation of surgical competence in the OR and result in prolongation of training. This unintended consequences of shift work and night float systems was described in a nationwide survey of US PGY1 residents as they are feeling less prepared to assume the PGY2 role.71 Simulation has emerged as an important adjunct to training for specific types of skills. It can prepare learners for laparoscopic, endoscopic, and endovascular procedures through use of geometric exercises, which can accustom them to the 2 dimensional environment, thus improving hand eye coordination and instrument handling. Well-designed simulation curricula for these types of procedures result in greater ease with routine parts of the operation and can decrease time to achieve competence and entry into practice.88 Simulation training also has an emerging role in teaching crisis resource management skills, such as leadership and communication;89 however, its impact on patient outcomes remains unknown. Simulation for open surgery is useful to assist residents to learn the repetitive parts of operations (eg, practicing a stapled bowel anastomosis using porcine material). These motor skills then become more “automatic,” allowing learners to focus on anatomy, pathology, and judgment when they are in the OR. However, relying on simulation to learn clinical reasoning, intraoperative decision-making, or refining sophisticated surgical techniques, especially for open surgery, is not feasible. Simulation models lack fidelity to human anatomy and tissues and are ill equipped to accommodate for variation, unexpected findings, and abnormal or diseased anatomy. Simulation training is best thought of as preparatory for residents' participation in the OR, as skills beyond the fundamental level in open surgery and accommodation to the 2 dimensional environment for laparoscopic procedures are better learned in the OR under the guidance and coaching of an experienced mentor.

tomy. Simulation training is best thought of as preparatory for residents' participation in the OR, as skills beyond the fundamental level in open surgery and accommodation to the 2 dimensional environment for laparoscopic procedures are better learned in the OR under the guidance and coaching of an experienced mentor. Perhaps most concerning is the trend toward increased failure rates on the oral component of the various surgical board examinations. Although correlation does not imply causation, as many factors are involved in the outcome of board examinations, the American Board of Thoracic Surgery reported that over the last decade there has been a nearly linear increase in the failure rate of candidates, with 35% of trainees being unsuccessful in 2012, compared with 12% in 2003.53 In General Surgery, the American Board failure rate has nearly doubled from 15% to 25% over 10 years, with a peak failure rate of 28% in 2011–2012.52 Interestingly, the results on written examination have remained static. Whereas performance on the written examination reflects knowledge and synthesis of information, performance on the oral examination tests judgement and candidates' real world experience with complex surgical issues. The American Board of Neurologic Surgeons has increased the length of residency training from 6 to 7 years, suggesting that ongoing RDH restrictions may affect the duration of training and time to achieve competence for independent practice. A survey conducted by the Fellowship Council of 145 subspecialty General Surgery program directors showed that they found starting fellows to be inadequately prepared for the OR; 38% demonstrated a lack of patient ownership, 30% could not independently perform a laparoscopic cholecystectomy, 66% were deemed unable to operate for 30 minutes unsupervised on a major procedure, and one-quarter were unable to recognize the early signs of complications.90

llows to be inadequately prepared for the OR; 38% demonstrated a lack of patient ownership, 30% could not independently perform a laparoscopic cholecystectomy, 66% were deemed unable to operate for 30 minutes unsupervised on a major procedure, and one-quarter were unable to recognize the early signs of complications.90 Trainees' involvement in emergency operations is important for the development of surgical judgement and competence. Residents learn from their integral involvement in patient care about the evolution of surgical diagnoses; when to operate, how much diagnostic accuracy is sufficient, and perhaps most importantly, to recognize and manage surgical complications. Major surgical complications are relatively infrequent, but they are serious. Rare presentations and infrequent complications of surgery require greater flexibility in resident scheduling, so that residents may benefit from these exposures, as expertise in these areas remains tied to experiences nested in the clinical environment. These aspects of surgical acumen are an essential part of professional maturation and require time on task. As emphasized in a recent commentary in the New England Journal of Medicine by Kesselheim et al. education and service are inextricably linked and the current focus on “duty hours” as an isolated consideration may undermine the overarching objective of postgraduate medical training.77 Our findings as well as others', emphasize that surgery perhaps more than other disciplines, requires immersion in the clinical environment, and that to a significant degree the training of a surgeon is well suited to the apprenticeship model.

may undermine the overarching objective of postgraduate medical training.77 Our findings as well as others', emphasize that surgery perhaps more than other disciplines, requires immersion in the clinical environment, and that to a significant degree the training of a surgeon is well suited to the apprenticeship model. 2011 ACGME Restrictions—16-Hour Duty Shift Maximums for Surgical Residents Review of the 17 studies that focused on assessing the impact of the 16-hour duty maximums showed a trend toward an overall negative impact in surgical training. Fargen et al65 showed in a survey of neurosurgical residents across the United States and Puerto Rico that 83% of respondents disagreed with, or strongly disagreed with 16-hour duty maximums for PGY1 residents. De Martino et al64 reported a similar concern among vascular surgery trainees. Recently, a national survey of 549 US program directors from Internal Medicine, Pediatrics and General Surgery programs revealed that 72% did not approve the 16-hour shift limit for interns. Although they generally did agree with other ACGME duty hour restrictions, they felt that currently, there were simply too many restrictions and that these restrictions resulted in negative effects on resident education, preparedness for senior roles, and no tangible benefits to resident wellness.70 These authors support greater flexibility with program and specialty specific tailoring of the educational environment for trainees. Our findings, and those of other authors, do not support 16-hour duty maximums for surgical residents at this time. Implementation of these RDH restrictions should be anticipated to result in prolongation of an already lengthy training duration and potential risk to patients, without tangible benefits to resident wellness.

findings, and those of other authors, do not support 16-hour duty maximums for surgical residents at this time. Implementation of these RDH restrictions should be anticipated to result in prolongation of an already lengthy training duration and potential risk to patients, without tangible benefits to resident wellness. Solutions to Improve Resident Wellness Streamlining Resident Workflow and Investment in Wellness Resources Residency systems have created expectations in teaching centers that the “doctor (aka resident) is always around.” This phenomenon dates back to the origins of the term “Resident.” Perhaps it is time to step back and reconsider the current role and purpose of surgical residency. Residents are present in teaching hospitals to learn about and deliver patient care. Their time and energy should be preserved to participate in patient evaluations, decision making, and operating. Surgical training needs to be more efficient. Other members of the health care team, such as physician assistants, nurse practitioners, respiratory therapists, pharmacists, phlebotomists, physiotherapists, and social workers, if correctly deployed could alleviate some of the “nonphysician” functions that currently distract residents from focusing their time in OR, management of ill patients, ambulatory clinics, academic rounds, and structured curricula.11,91 This approach would improve residents' professional satisfaction and sense of wellness. Additional strategies to alleviate fatigue and prevent burnout include improved opportunities for sleep while on call as was recommended in the 2009 Institute of Medicine report. These solutions more readily respect circadian physiology than the current trend toward shift work and night float systems.92,93 Improved handover tools, innovations such as electronic rounding reports, and polices to escalate care to more senior residents and faculty, when appropriate, would alleviate resident workload and improve patient safety.94

spect circadian physiology than the current trend toward shift work and night float systems.92,93 Improved handover tools, innovations such as electronic rounding reports, and polices to escalate care to more senior residents and faculty, when appropriate, would alleviate resident workload and improve patient safety.94 Expansion of OR resources and acute care surgery teams may be necessary in certain settings to better ensure urgent operative cases are completed during daytime hours when feasible, leaving overnight OR resources for patients who present during the evening and night with life, limb, or vision threatening surgical diagnoses.95,96 There will remain a subset of patients who require urgent operative intervention after midnight. Resident participation in these patients' preoperative and operative care is essential for residents' professionalization.

present during the evening and night with life, limb, or vision threatening surgical diagnoses.95,96 There will remain a subset of patients who require urgent operative intervention after midnight. Resident participation in these patients' preoperative and operative care is essential for residents' professionalization. Expansion of Wellness Resources and Implementation of Wellness Programs Physicians, and perhaps to a greater extent surgeons, have long ignored our own health. This phenomenon in part is related to our professionalization and societal expectations, but it is also connected to the reality that every single one of our patients at any given moment is much worse off than the treating doctor. For the resident on a ward of cancer, trauma, sepsis, or burn patients, prioritizing one's own health is nearly impossible. Structural changes are required to accommodate this reality and enable residents to attend to their own physical and emotional well-being when necessary. One of the key recommendations in the Royal College of Physicians and Surgeons of Canada National Steering Committee report is to move the current discussion away from one that is focused solely on RDH and includes institutional strategies that enable fatigue mitigation and management, rather than relying only on externally imposed RDH limits.5 This strategy, in part, could include wellness programs within residency structures that screen for and are equipped to manage psychological distress and burnout. Wellness programs are defined by a combination of active and passive initiatives targeting the various domains of physical, mental, social, and intellectual wellness and have been shown to be effective, inexpensive, and well received by trainees.97–99

n for and are equipped to manage psychological distress and burnout. Wellness programs are defined by a combination of active and passive initiatives targeting the various domains of physical, mental, social, and intellectual wellness and have been shown to be effective, inexpensive, and well received by trainees.97–99 CONCLUSIONS Systematic review of the literature related to RDH restrictions in surgery shows that there were benefits to resident wellness realized from 2003 ACGME regulations limiting RDH to 80 hours per week. However, there seems to be some concern for increased morbidity and mortality rates in high-acuity patients as a result of the 2003 and to a greater extent the 2011 ACGME regulations.11,12,20 Also concerning is the trend for poorer performance on American Board of Surgery certification examinations over the last decade. Further evaluation of the literature related specifically to the implementation of 16-hour duty maximums for PGY1 residents in the United States and all in-house residents in the province of Quebec showed that there may be little added benefit to residents' wellness.67 Recent evidence suggests that adverse patient outcomes may extend beyond the highest acuity patients in surgery, as residents perceive increased rates of medical errors as a result of 16-hour duty maximums. Equally concerning, we found no evidence for improved educational or training experiences as a result of 16-hour duty maximums. Rather, there seems to be consistent findings across many studies of increased patient handovers, poorer team integration, professional dissatisfaction, concern for maturation of clinical skills, and decreased time in the OR. Philibert et al100 recently published a literature review of high-quality articles with similar conclusions.

seems to be consistent findings across many studies of increased patient handovers, poorer team integration, professional dissatisfaction, concern for maturation of clinical skills, and decreased time in the OR. Philibert et al100 recently published a literature review of high-quality articles with similar conclusions. Fatigue is potentially harmful, mitigating its effects while ensuring adequate training is essential. Strategies such as less than 24 hour shifts have failed to demonstrate a significant benefit to resident wellness and potentially deleterious consequences for patient and training outcomes.35,38,55 This observation has also been documented in critically ill patients in the ICU setting.81 Training should prepare residents for schedules commonly encountered in the independent practice of that specialty. It is clear that a “one-size fits all” solution to the issue of RDH restrictions is not appropriate for all medical disciplines. A more tailored and discipline specific approach is required such that training objectives and patient outcomes are protected. This view has been supported by a pan-Canadian consensus process related to RDHs, led by the Royal College of Physicians and Surgeons in Canada.5 This report is the first nonsurgical document to support this perspective, one that has been emphasized by many surgical education bodies including the American College of Surgeons, Division of Education and the Royal College of Surgeons of Edinburgh.3,101 The challenge for surgical training programs will be how best to support resident wellness though fatigue mitigation strategies, screening and addressing resident burnout, as well as improving resident workflow to optimize resident sleep, supervision and minimize tasks unrelated to the provision of medically necessary care, and preserving access to key mentorship and high impact learning opportunities. An approach that centers on the training mandate of residencies and allows a degree of flexibility will be necessary if we wish to preserve robust educational and patient-level outcomes without further prolonging an already arduous and lengthy training experience. It is well time to reconsider the direction of RDH as it relates to surgical training, as the recent erosion of resident training time has not achieved the desired results. A broader dialogue beyond duty hours alone will yield a more satisfactory outcome for all stakeholders, and especially for surgical residents.

nce. It is well time to reconsider the direction of RDH as it relates to surgical training, as the recent erosion of resident training time has not achieved the desired results. A broader dialogue beyond duty hours alone will yield a more satisfactory outcome for all stakeholders, and especially for surgical residents. ACKNOWLEDGMENTS The authors thank the National Steering Committee on RDHs and the Royal College of Physicians and Surgeons of Canada for their assistance in the completion of this project and especially Lisa Gorman and Sarah Taber. They also thank the following working group members for their inputs: Barbara Guido, Henry Broekhuyse, Joanne Carrier, David Steven, Douglas Hedden, Jefferson Wilson, Jean Francios Chevalier, Joel Werier, Pamela Chu, Piotr Blachut, Ravi Sidhu, Grace Yeung, and Mark Walsh. Disclosure: This study was supported by Health Canada and the Department of Surgery, University of Toronto, Ontario, Canada. The authors declare that they have no conflict of interest.

Esophagectomy is associated with a high morbidity rate, in large part related to the reconstruction. Typically reconstruction is done with a tubularized gastric graft perfused by the right gastroepiploic arcade. Perfusion at the proximal portion of the graft, in the area where the anastomosis is typically created, is variable but often tenuous because the gastroepiploic arcade rarely reaches the tip of the graft. Instead, the most proximal portion of the graft is typically perfused by intramural capillaries within the wall of the stomach and small vessels in the omentum along the greater curvature. It has been shown that cervical esophagogastric anastomotic complications are increased in patients with a history of cigarette smoking, neoadjuvant radiotherapy to the fundus of the stomach, hypertension, and diabetes.1 These factors may affect perfusion in the area of the proposed anastomosis by interfering with microvascular integrity.

at cervical esophagogastric anastomotic complications are increased in patients with a history of cigarette smoking, neoadjuvant radiotherapy to the fundus of the stomach, hypertension, and diabetes.1 These factors may affect perfusion in the area of the proposed anastomosis by interfering with microvascular integrity. Leak rates after a cervical esophagogastrostomy are among the highest of any gastrointestinal anastomosis.2 Compromised microperfusion in the area where the anastomosis is sited likely contributes to leaks, but until recently there has not been an easy and reliable technique to intraoperatively evaluate microperfusion. Laser-assisted angiography (LAA) with the SPY Imaging System (Novadaq, Ontario, Canada) provides real-time intraoperative assessment of both gross and microperfusion. The SPY system was developed for cardiac surgery but has been applied in plastic surgery to evaluate perfusion in flaps and tissue transfers.3–5 We hypothesized that this technology might provide valuable real-time information about perfusion in a gastric graft during an esophagectomy with reconstruction. The aim of the study was to review our experience with the SPY system during esophagectomy and gastric pull-up (GPU) and to correlate our assessment of perfusion at the proposed site of the anastomosis with subsequent anastomotic healing and leak rates.

on in a gastric graft during an esophagectomy with reconstruction. The aim of the study was to review our experience with the SPY system during esophagectomy and gastric pull-up (GPU) and to correlate our assessment of perfusion at the proposed site of the anastomosis with subsequent anastomotic healing and leak rates. METHODS After initially using intraoperative LAA on several esophagectomies with GPU we saw that perfusion in the tip of the gastric graft as assessed visibly with the SPY system was variable. In some patients, the entire graft showed rapid, bright perfusion by LAA (Fig. 1). Often, though, there was a point of demarcation where rapid and bright perfusion transitioned into slower and less robust perfusion. We prospectively decided to evaluate the utility of LAA by placing a suture at the site of this transition if present, and recording in the operative note the location of the anastomosis relative to the suture (proximal vs distal to the stitch) (Fig. 2). During this time, we used standard techniques including our clinical judgment and Doppler to evaluate graft perfusion for placement of the anastomosis. In addition to noting the site of a transition in perfusion by LAA if present, we also noted the proximal-most extent of the Doppler signal. We subsequently evaluated the outcome of these anastomoses by retrospectively reviewing the records of these patients. The study was approved by the University of Southern California institutional review board. FIGURE 1. Rapid and bright perfusion of the entire gastric graft by LAA.

METHODS After initially using intraoperative LAA on several esophagectomies with GPU we saw that perfusion in the tip of the gastric graft as assessed visibly with the SPY system was variable. In some patients, the entire graft showed rapid, bright perfusion by LAA (Fig. 1). Often, though, there was a point of demarcation where rapid and bright perfusion transitioned into slower and less robust perfusion. We prospectively decided to evaluate the utility of LAA by placing a suture at the site of this transition if present, and recording in the operative note the location of the anastomosis relative to the suture (proximal vs distal to the stitch) (Fig. 2). During this time, we used standard techniques including our clinical judgment and Doppler to evaluate graft perfusion for placement of the anastomosis. In addition to noting the site of a transition in perfusion by LAA if present, we also noted the proximal-most extent of the Doppler signal. We subsequently evaluated the outcome of these anastomoses by retrospectively reviewing the records of these patients. The study was approved by the University of Southern California institutional review board. FIGURE 1. Rapid and bright perfusion of the entire gastric graft by LAA. FIGURE 2. A transition point is seen between rapid and bright versus slower, less robust perfusion (arrow). A suture is placed at the site of this transition point and if possible the anastomosis was placed proximal to this suture in an area of good perfusion by LAA.

FIGURE 1. Rapid and bright perfusion of the entire gastric graft by LAA. FIGURE 2. A transition point is seen between rapid and bright versus slower, less robust perfusion (arrow). A suture is placed at the site of this transition point and if possible the anastomosis was placed proximal to this suture in an area of good perfusion by LAA. Laser-assisted Indocyanine Green Fluorescent-Dye Angiography The SPY Imaging System is designed to acquire fluorescence images. This is achieved using a fluorescent agent, indocyanine green (ICG), which upon intravascular administration is rapidly and extensively (>98%) bound to plasma proteins. The ICG is a sterile, water soluble, tricarbocyanine dye with a peak spectral absorption at 800 to 810 nm in blood. It is taken up by the liver and excreted unchanged into the bile. The plasma half-life of ICG is 3 to 5 minutes. The ICG is caused to fluoresce by illumination with a laser at 806 nm. The ICG molecules absorb the light, get “excited,” drop back to an “unexcited” state, and emit light in a longer wavelength. Fluorescence images of the GPU graft are acquired using a charge coupled device video camera, sensitive into the near infrared, and equipped with an edge filter to permit efficient transmission of fluorescent light (in the range 815–880 nm) while blocking laser and room light. The images are visible in real-time on a monitor and are captured to hard drive for analysis, future review, and archiving.

ideo camera, sensitive into the near infrared, and equipped with an edge filter to permit efficient transmission of fluorescent light (in the range 815–880 nm) while blocking laser and room light. The images are visible in real-time on a monitor and are captured to hard drive for analysis, future review, and archiving. Safety of Indocyanine Green Indocyanine green (ICG) has been approved for use in humans and has been extensively used for determining cardiac output, hepatic function, liver blood flow, for ophthalmologic angiography, and during cardiac bypass and plastic and reconstructive surgery. The incidence of adverse reactions to ICG is low and most are mild (sore throat, feeling of warmth). Rare reports describe hypotension requiring treatment with epinephrine. Caution is recommended in patients with iodine or shellfish allergy. Surgical Procedure All patients underwent esophagectomy with a transhiatal, en bloc transthoracic, minimally invasive thoracoscopic/laparoscopic or vagal-sparing approach. The tubularized gastric graft was brought up through the posterior mediastinum and a handsewn single-layer cervical anastomosis onto the anterior wall of the gastric pull-up was performed in all patients.

with a transhiatal, en bloc transthoracic, minimally invasive thoracoscopic/laparoscopic or vagal-sparing approach. The tubularized gastric graft was brought up through the posterior mediastinum and a handsewn single-layer cervical anastomosis onto the anterior wall of the gastric pull-up was performed in all patients. Intraoperative Assessment of Graft Perfusion and Placement of the Anastomosis After creating a 3- to 4-cm wide gastric tube, but before bringing the graft up to the neck, the SPY system was used to assess graft perfusion. The time between gastric tube creation and LAA was approximately 15 minutes. Images were obtained beginning 5 seconds after intravenous central-line injection of 2.5 mg of ICG followed by a 5 mL flush of saline. This dose was recommended by Novadaq and provided excellent visualization of microperfusion in the gastric grafts. After LAA assessment of perfusion and placement of a suture, if a transition point was seen, the graft was brought up to the neck through the posterior mediastinum. The anastomosis was placed in a suitable area of the anterior wall of the gastric graft, and the location of the anastomosis relative to the suture, if present, was recorded in the operative note. An effort was always made to place the anastomosis proximal to the suture, in an area of good perfusion by LAA, but in patients where the anastomosis would not reach to a portion of stomach proximal to the suture it was placed at or distal to the suture.

ve to the suture, if present, was recorded in the operative note. An effort was always made to place the anastomosis proximal to the suture, in an area of good perfusion by LAA, but in patients where the anastomosis would not reach to a portion of stomach proximal to the suture it was placed at or distal to the suture. Assessment of Anastomotic Healing A videoesophagram was routinely obtained at 5 to 7 days postoperatively, and if there was evidence of a leak or abnormality, or when indicated by clinical deterioration of the patient, upper endoscopy was performed. If a leak was evident either by videoesophagram or by upper endoscopy, the patient was classified as having an anastomotic leak. Furthermore, for this study, the leak was defined as minor when conservative treatment with antibiotics or nil per os was sufficient to lead to healing and major when an intervention (endoscopic stenting) or reoperation was necessary. Comorbidities and risk factors for poor anastomotic healing that were evaluated included cardiac disease, hypertension, diabetes, chronic obstructive pulmonary disease, current cigarette smoking, or alcohol abuse. Statistical Analysis Data are expressed as medians and interquartile range. Comparisons of proportions were performed using the Fisher exact test. A P < 0.05 was considered significant. Univariate and multivariate analysis were performed using Prism software (GraphPad, La Jolla, CA).

Comorbidities and risk factors for poor anastomotic healing that were evaluated included cardiac disease, hypertension, diabetes, chronic obstructive pulmonary disease, current cigarette smoking, or alcohol abuse. Statistical Analysis Data are expressed as medians and interquartile range. Comparisons of proportions were performed using the Fisher exact test. A P < 0.05 was considered significant. Univariate and multivariate analysis were performed using Prism software (GraphPad, La Jolla, CA). RESULTS There were 150 consecutive patients that had intraoperative LAA assessment of perfusion during esophagectomy with GPU using the SPY system from March 2008 until July 2011. The median age of the patients was 66.7 years (interquartile range: 57–74). There were 125 men and 25 women. The indication for esophagectomy was cancer in 133 patients and end-stage benign disease in 17 patients. The type of esophagectomy was en bloc in 88 patients, transhiatal in 26, minimally invasive in 24 and vagal-sparing in 12 (open 10, laparoscopic 2). Major complications occurred in 22% and there was one mortality (Table 1). The median hospital stay in all patients was 14 days. The median hospital stay was significantly longer in patients with a leak (20 days with leak vs 13 days with no leak; P = 0.0096) and in those with a major versus a minor leak (40.5 days with a major leak vs 18 days with a minor leak, P = 0.0198).

rtality (Table 1). The median hospital stay in all patients was 14 days. The median hospital stay was significantly longer in patients with a leak (20 days with leak vs 13 days with no leak; P = 0.0096) and in those with a major versus a minor leak (40.5 days with a major leak vs 18 days with a minor leak, P = 0.0198). TABLE 1. Minor and Major Complications (n = 150) 52 (34.7%) Minor complications Wound infection 10 (6.7%) Wound hematoma, dehiscence 2 (1.3%) Urinary tract infection 2 (1.3%) Atrial fibrillation 24 (16%) Pulmonary complications 13 (8.6%) Deep venous thrombosis 1 (0.7%) Major complications 33 (22%) Perforation with NG tube 1 (0.7%) Leakage (minor/major) 24 (16/8) Chylothorax 7 (4.6%) Pulmonary Embolism 1 (0.7%) Mortality 1 (0.7%) Intraoperative injection of ICG was well-tolerated by all patients with no adverse events or noticeable adverse effects. There were no technical difficulties and images were obtained in all patients. The entire graft was noted to have good perfusion in 66 of 150 patients (44%), whereas in 84 (66%) patients a line of demarcation was noted between rapid, bright perfusion and slower, less robust perfusion in the fundus of the gastric tube. In these 84 patients, the anastomosis was placed proximal to the stitch in 29 patients, at or distal to the stitch in 49 patients and in 6 patients no anastomosis was performed. These 6 patients all had significant comorbid conditions and poor perfusion by LAA in the area where the anastomosis was to be performed, and we elected to delay the reconstruction until ischemic conditioning led to better graft perfusion. In these patients, the graft was brought up to the neck and sutured to the sternocleidomastoid muscle as has been previously described.6 Subsequent reconstruction was done once the patient had satisfactorily recovered from the esophagectomy, usually at about 8 to 12 weeks.

ic conditioning led to better graft perfusion. In these patients, the graft was brought up to the neck and sutured to the sternocleidomastoid muscle as has been previously described.6 Subsequent reconstruction was done once the patient had satisfactorily recovered from the esophagectomy, usually at about 8 to 12 weeks. Anastomotic leaks occurred in 24 (16.7%) of the 144 patients who had an anastomosis and were classified as major in 8 and minor in 16 patients. Patients in whom the anastomosis was placed at or distal to the site of the suture were significantly more likely to have a leak compared with those in whom no suture was placed (entire graft well-perfused) or where the anastomosis was placed proximal to the transition point in an area of good perfusion by LAA (45% vs 2%, P < 0.0001) (Fig. 3). Similarly, major leaks were significantly more common when the anastomosis was not placed in an area of good perfusion by LAA (15% vs 0; P = 0.0002). Major leaks were treated with an endoscopic stent in 4 patients and reoperation with neck drainage in 4 patients. No patient required graft takedown. All leaks except 1 occurred in patients with cancer, and all patients with leaks had at least 1 comorbid condition. On univariate analysis, placement of the anastomosis at or distal to the suture and a history of hypertension were significantly associated with an anastomotic leak (Table 2). By multivariate analysis, anastomotic placement at or distal to the suture was the only significant factor associated with a leak.

id condition. On univariate analysis, placement of the anastomosis at or distal to the suture and a history of hypertension were significantly associated with an anastomotic leak (Table 2). By multivariate analysis, anastomotic placement at or distal to the suture was the only significant factor associated with a leak. FIGURE 3. The frequency of major and minor leaks in patients where the anastomosis was placed in an area of good perfusion by LAA [either entire graft with good perfusion (n = 66) or anastomosis placed proximal to suture at site of transition (n = 29)] versus those patients that had an anastomosis placed at or distal to the site of the suture at the transition point (n = 49) (2% vs 45%, P < 0.0001).

was placed in an area of good perfusion by LAA [either entire graft with good perfusion (n = 66) or anastomosis placed proximal to suture at site of transition (n = 29)] versus those patients that had an anastomosis placed at or distal to the site of the suture at the transition point (n = 49) (2% vs 45%, P < 0.0001). TABLE 2. Univariate Analysis of Factors Potentially Associated With Leak (n = 144) No Leak (n = 120) Leak (n = 24) P Anastomosis at or distal to suture 27 22 <0.0001 Cancer diagnosis 104 23 0.31 Diabetes 23 5 0.78 Hypertension 57 18 0.0149 Coronary artery disease 25 3 0.41 Active or recent smoking 51 14 0.18 Neoadjuvant therapy 52 15 0.12 DISCUSSION Adequate perfusion is a prerequisite for reliable healing of a gastrointestinal anastomosis. One of the most tenuous anastomoses in all of gastrointestinal surgery is the cervical esophagogastric anastomosis during esophagectomy with gastric reconstruction. Leaks are reported in 20% to 35% of these patients and are a major source of short- and long-term morbidity and occasionally mortality.7 Intraoperative assessment of gastric graft perfusion has typically been on the basis of color, temperature, and Doppler signal. A bluish color and cool temperature are unsettling but lack specificity, and although the Doppler is good for gross perfusion, it is unreliable for microperfusion. Seldom is there a discernible Doppler signal beyond about two thirds the way up the greater curvature of the gastric tube and thus the Doppler is not useful to assess perfusion in the area where the anastomosis is likely to be performed in most patients. In this study, we found that the Doppler signal always disappeared proximal to the site of demarcation by LAA (if present) and could not differentiate grafts with complete microperfusion to the tip versus those with a zone of demarcation as seen using the SPY technology (data not shown).

be performed in most patients. In this study, we found that the Doppler signal always disappeared proximal to the site of demarcation by LAA (if present) and could not differentiate grafts with complete microperfusion to the tip versus those with a zone of demarcation as seen using the SPY technology (data not shown). We started using the SPY system to see if it would provide real-time information about perfusion, particularly microperfusion, that could be useful to assess gastric grafts during esophagectomy. Because we were uncertain of the significance of perfusion as visually assessed with the SPY system, we decided to place a stitch if there was a transition by LAA between fast and bright versus slower and less robust perfusion. We found that an anastomosis was unlikely to leak in patients with no transition point or when it was placed proximal to the transition point when present, in an area of good LAA perfusion. The leaks that occurred in these patients were all minor and healed with conservative therapy. In contrast, 45% of patients with an anastomosis placed at or distal to the stitch, in an area of slower and less robust perfusion by LAA, had an anastomotic leak. Furthermore, 36% of these leaks were major leaks that required an intervention.

urred in these patients were all minor and healed with conservative therapy. In contrast, 45% of patients with an anastomosis placed at or distal to the stitch, in an area of slower and less robust perfusion by LAA, had an anastomotic leak. Furthermore, 36% of these leaks were major leaks that required an intervention. There are several important issues in relation to this finding. First, when perfusion to the site of the planned anastomosis was good by LAA leaks were unlikely and minor if they occurred. These patients may be considered for fast tracking or elimination of a barium swallow in the absence of any clinical suggestion of a leak. In contrast, when the anastomosis had to be placed at or distal to a transition point in perfusion by LAA, a leak will develop in almost one-half of the patients, and some of these leaks will require an intervention. These patients should be monitored closely and evaluated promptly for any evidence of clinical deterioration. Furthermore, in these patients, planned graft evaluation with an upper endoscopy at 5 to 7 days after reconstruction may be useful to address a leak before it becomes clinically significant. We have previously shown the safety and efficacy of early endoscopy after esophagectomy and reconstruction.8 It is important to recognize that in these patients with an anastomosis at or distal to the transition site healing without leak occurred in more than half of the patients. Therefore, when faced with this situation factors such as the patient's ability to tolerate a leak should be taken into consideration.

nstruction.8 It is important to recognize that in these patients with an anastomosis at or distal to the transition site healing without leak occurred in more than half of the patients. Therefore, when faced with this situation factors such as the patient's ability to tolerate a leak should be taken into consideration. If there were no intraoperative options then defining perfusion with the SPY system would be informative for postoperative care but would not be clinically useful at the time of the operation. However, since understanding the implications of putting the anastomosis in an area of less robust perfusion, we now will alter our operative plan in patients at high risk for doing poorly with a leak. These patients include those with significant comorbid conditions or very elderly patients who tend to have little reserve for major complications. In this series, we had 6 such patients in whom we elected to delay the reconstruction. With this technique, the graft is left in the neck but no anastomosis is performed. Over several weeks, ischemic conditioning leads to improved perfusion in the graft and the subsequent anastomosis typically heals reliably in these patients.6,9 Alternatively, in some cases, we will now resect a portion of the manubrium and first rib and place the anastomosis proximal to the stitch either with the graft in the posterior mediastinum or in a substernal location. Using this strategy, we can assess the risk for anastomotic leak and tailor the operative plan as necessary when the perfusion by LAA or condition of the patient indicates it appropriate to do so.

lace the anastomosis proximal to the stitch either with the graft in the posterior mediastinum or in a substernal location. Using this strategy, we can assess the risk for anastomotic leak and tailor the operative plan as necessary when the perfusion by LAA or condition of the patient indicates it appropriate to do so. In an effort to further refine assessment of perfusion and to move from qualitative to quantitative assessment, the SPY-Q system has been introduced (Fig. 4). Efforts are underway to define a threshold of perfusion below which the majority of anastomoses will not heal. In this way, the risk of anastomotic leak can be defined even more precisely and operative decisions tailored more specifically for an individual patient. Since gaining confidence with the SPY system, we have expanded use of LAA to all types of reconstruction after esophagectomy including colon and jejunal grafts and have found it equally useful to evaluate perfusion in these grafts. Furthermore, the recent introduction of the Pinpoint (Novadaq Ontario, Canada) and Firefly (Intuitive Surgical, Sunnyvale, CA) systems allow LAA perfusion assessment during minimally invasive and robotic procedures. FIGURE 4. Perfusion by LAA shown in (A) qualitative mode versus (B) with quantitative (SPY-Q) overlay.

In an effort to further refine assessment of perfusion and to move from qualitative to quantitative assessment, the SPY-Q system has been introduced (Fig. 4). Efforts are underway to define a threshold of perfusion below which the majority of anastomoses will not heal. In this way, the risk of anastomotic leak can be defined even more precisely and operative decisions tailored more specifically for an individual patient. Since gaining confidence with the SPY system, we have expanded use of LAA to all types of reconstruction after esophagectomy including colon and jejunal grafts and have found it equally useful to evaluate perfusion in these grafts. Furthermore, the recent introduction of the Pinpoint (Novadaq Ontario, Canada) and Firefly (Intuitive Surgical, Sunnyvale, CA) systems allow LAA perfusion assessment during minimally invasive and robotic procedures. FIGURE 4. Perfusion by LAA shown in (A) qualitative mode versus (B) with quantitative (SPY-Q) overlay. A limitation of this study is that the evaluation of the SPY images is largely qualitative at this point, and as we have gained experience with the images and perfusion implications we undoubtedly altered our practice, which may have impacted the results of this retrospective review. However, if anything these alterations would likely have led to a reduced rate of leaks since with experience, we began making even greater efforts to place the anastomosis proximal to the suture or alter the surgical plan. Furthermore, our overall small number of leaks prohibited an evaluation of the impact of operative approach on anastomotic healing and likely masked the role of important comorbid conditions that contribute to leaks.

making even greater efforts to place the anastomosis proximal to the suture or alter the surgical plan. Furthermore, our overall small number of leaks prohibited an evaluation of the impact of operative approach on anastomotic healing and likely masked the role of important comorbid conditions that contribute to leaks. CONCLUSIONS Perfusion is critical for anastomotic healing after esophagectomy and gastric pull-up. Intraoperative real-time assessment of perfusion with LAA correlated with the likelihood of an anastomotic leak, and the cervical anastomotic leak rate was trivial when the anastomosis was placed in an area of the graft shown to have good perfusion. The use of LAA during esophagectomy with gastric pull-up may lead to an altered surgical plan in some patients and contribute to reduced anastomotic morbidity and better overall patient outcomes. Presented as an oral presentation at the EAES Meeting on June 20, 2013, Vienna, Austria. Disclosure: S.R.D. is a consultant for Novadaq Technologies Inc, which manufactures the SPY device. The authors declare no conflicts of interest.

The collateral damage to the abdominal wall for intraperitoneal access accounts for a major part of the operative trauma in abdominal surgery. It is also a source of many intra- and postoperative complications. Minimally invasive surgery was developed to minimize this damage. Laparoscopic cholecystectomy (LC) using two 10-mm and two 5-mm trocars became the gold standard many years ago, when its advantages in terms of postoperative pain, aesthetic results, and duration of hospital stay compared with conventional open cholecystectomy were shown.1–4 Using instruments with a lower diameter for a “needlescopic cholecystectomy” (NC), which was evaluated in several prospective, randomized, sometimes even double-blinded trials, further reduced postoperative pain and improved aesthetics.5,6 The remaining 10-mm trocar incision, which is also used for retrieval, emerged to be the most painful incision.7 However, this abdominal wall trauma for instrument access and specimen retrieval can be avoided by accessing the abdominal cavity through natural orifices, for example, transvaginally via the posterior vault of the vagina [natural orifice transluminal endoscopic surgery (NOTES)]. Admittedly, because of limitations of avoidable instruments, NOTES cholecystectomy has only been introduced into clinical routine as a hybrid procedure [transvaginal cholecystectomy (TVC)].8 It facilitates an additional transumbilical 5-mm trocar resulting in TVC, as described by Zornig et al.9 Our prospective, randomized trial was designed to evaluate TVC in terms of lesser postoperative pain, which was so far only shown in nonrandomized trials.10–13 Even if the advantages of NC against the traditional cholecystectomy are limited, TVC was still supposed to be compared with the least invasive laparoscopic technique. Thus, patients undergoing NC were studied as a control group. The emphasis was put on postoperative pain intensity and the safety of transvaginal access.

ven if the advantages of NC against the traditional cholecystectomy are limited, TVC was still supposed to be compared with the least invasive laparoscopic technique. Thus, patients undergoing NC were studied as a control group. The emphasis was put on postoperative pain intensity and the safety of transvaginal access. METHODS Study Design and Patients The needlescopic versus transvaginal cholecystectomy study was a randomized, prospective, single-center, and nonblinded clinical trial comparing transvaginal/transumbilical cholecystectomy with 3-trocar NC. Between February 2010 and June 2012, eligible patients were recruited in the Department of Abdominal, Vascular and Transplant Surgery of the Cologne-Merheim Medical Center. The protocol was approved by the Research Ethics Committee of the Witten/Herdecke University (89/2009). Written and informed consent was obtained from all patients. Inclusion criteria were as follows: female sex, indication for elective cholecystectomy because of symptomatic cholecystolithiasis, age between 18 and 80 years, and legal competence.

METHODS Study Design and Patients The needlescopic versus transvaginal cholecystectomy study was a randomized, prospective, single-center, and nonblinded clinical trial comparing transvaginal/transumbilical cholecystectomy with 3-trocar NC. Between February 2010 and June 2012, eligible patients were recruited in the Department of Abdominal, Vascular and Transplant Surgery of the Cologne-Merheim Medical Center. The protocol was approved by the Research Ethics Committee of the Witten/Herdecke University (89/2009). Written and informed consent was obtained from all patients. Inclusion criteria were as follows: female sex, indication for elective cholecystectomy because of symptomatic cholecystolithiasis, age between 18 and 80 years, and legal competence. Exclusion criteria were as follows: acute cholecystitis or locally complicated disease (eg, gallbladder empyema, choledocholithiasis, and pancreatitis), liver cirrhosis (Child–Pugh A, B, and C), severe comorbidity, class IV or V as defined by the American Society for Anesthesiologists (ASA), previous malignancy or suspected malignancy in preoperative imaging, a body mass index (BMI) higher than 40 kg/m2, chronic abuse of analgesics or alcohol, neuromuscular disease that could interfere with treatment or measures of pain, history of major abdominal surgery with a high risk of intraperitoneal adhesions (minor operations such as an appendectomy, inguinal hernia repair, and minor gynecological surgery were not considered exclusion criteria), gravidity or breast-feeding, allergy against analgesics, patients who are dependent on or employed by the trial sponsor or physicians, participation in other clinical studies that could interfere with the present trial, and no written informed consent signed.

necological surgery were not considered exclusion criteria), gravidity or breast-feeding, allergy against analgesics, patients who are dependent on or employed by the trial sponsor or physicians, participation in other clinical studies that could interfere with the present trial, and no written informed consent signed. Enrolled patients were randomly assigned on a 1:1 ratio to either a TVC or an NC. Randomization of the 40 patients was conducted without stratification using numbered, opaque, sealed envelopes.14 Computer-aided randomization of numbers in randomly lined up blocks of 4 or 6 was generated by an independent statistician.

necological surgery were not considered exclusion criteria), gravidity or breast-feeding, allergy against analgesics, patients who are dependent on or employed by the trial sponsor or physicians, participation in other clinical studies that could interfere with the present trial, and no written informed consent signed. Enrolled patients were randomly assigned on a 1:1 ratio to either a TVC or an NC. Randomization of the 40 patients was conducted without stratification using numbered, opaque, sealed envelopes.14 Computer-aided randomization of numbers in randomly lined up blocks of 4 or 6 was generated by an independent statistician. Diagnosis and Postoperative Course A thorough history taking and physical examination, sonography, and laboratory values were obtained preoperatively for all patients. A gynecologist examined all patients randomized to the TVC group using a standardized examination form on the day before the operation. Therefore, the sealed envelopes were opened on the preoperative day for randomization, and thus the patients were informed in which group they were. Contraindications for a transvaginal procedure included were as follows: lacking visibility of the cervix; ongoing pregnancy; genital infections; known endometriosis; neoplasms of the vulva, vagina, or cervix; and intact hymen. These criteria did not result in the exclusion of any patients from the TVC procedure. In both groups, a single-shot antibiotic (cefuroxime 1.5 g) was intravenously administered preoperatively. Before every skin incision, a local anesthetic (3 mL of bupivacaine 0.25%) was administered subcutaneously in both groups. Postoperative analgesia in the recovery room was administered only on demand with a peripheral (paracetamol) or a centrally acting analgesic (piritramide, an opioid with a morphine-equivalence factor of 0.7). Postoperative pain medication was standardized and identical for both groups. The following analgesics were offered on the ward: on the day of surgery, 2 × 1 g of paracetamol (Perfalgan) intravenously; on postoperative day (POD) 1, 3 × 2 tablets of paracetamol 500 mg; and from POD 2, 3 × 1 tablet of paracetamol 500 mg. On demand, 7.5 mg of piritramide (Dipidolor) was administered subcutaneously or in a 100-mL short-infusion intravenously. According to hospital standards, low-molecular-weight heparin (nadroparine) was used for thrombosis prophylaxis during the hospital stay. Full oral intake and mobilization were regularly begun on the day of surgery. On POD 2, laboratory values were taken. Regularly on POD 2 or earlier on demand, bandages were taken off and the first wound examination took place. In case of clinical symptoms or atypical laboratory results, an abdominal sonogram was obtained.

l oral intake and mobilization were regularly begun on the day of surgery. On POD 2, laboratory values were taken. Regularly on POD 2 or earlier on demand, bandages were taken off and the first wound examination took place. In case of clinical symptoms or atypical laboratory results, an abdominal sonogram was obtained. Patients were dismissed from POD 2 under the following conditions: complete oral intake, subjective well-being, primary healing of wounds, regular laboratory results, and if applicable, a normal sonogram. All patients were examined in the first ambulatory posthospital checkup on POD 10. On this occasion, the pain and analgesic diary was collected and evaluated. The patient was examined and questioned. The aesthetic result was evaluated by both the examining physician and the patient using an ordinal scale for satisfaction. Patients from the TVC group were advised against penetrating sexual intercourse for 2 weeks postoperatively. Also, they were examined by a gynecologist 12 days after surgery, again using a standardized examination form.

Patients were dismissed from POD 2 under the following conditions: complete oral intake, subjective well-being, primary healing of wounds, regular laboratory results, and if applicable, a normal sonogram. All patients were examined in the first ambulatory posthospital checkup on POD 10. On this occasion, the pain and analgesic diary was collected and evaluated. The patient was examined and questioned. The aesthetic result was evaluated by both the examining physician and the patient using an ordinal scale for satisfaction. Patients from the TVC group were advised against penetrating sexual intercourse for 2 weeks postoperatively. Also, they were examined by a gynecologist 12 days after surgery, again using a standardized examination form. Surgical Technique The transvaginal/transumbilical cholecystectomy was performed with rigid reusable instruments in the lithotomy position, as described by Zornig et al.9 The first surgeon was standing on the left of the patient, the second surgeon between the legs. A 12-mm Hg capnoperitoneum was established via an umbilical Veress needle. An umbilical 6-mm trocar (Karl Storz GmbH & Co KG, Tuttlingen, Germany), a transvaginal, curved 5-mm grasping forceps (according to CUSCHIERI O-CON, 43 cm long, Karl Storz GmbH & Co KG, Tuttlingen, Germany), and a transvaginal 11-mm trocar without connector for insufflation (Karl Storz GmbH & Co KG, Tuttlingen, Germany), which were inserted via the posterior vault of the vagina, were used. The dissection of the gallbladder, the cystic duct, and the cystic artery, and clipping (Endo Clip 5-mm clip applier, Covidien, MA) and transecting of them, was done via the umbilical 6-mm trocar while viewing through a transvaginal 10-mm optic (45 degrees, 42 cm long, Karl Storz GmbH & Co KG, Tuttlingen, Germany). The gallbladder was transvaginally extricated through the 11-mm trocar incision in the posterior vault after changing the view to a transumbilical 5-mm optic (45 degrees, 29 cm long, Karl Storz GmbH & Co KG, Tuttlingen, Germany). The 2 small incisions in the posterior vault were closed with resorbable sutures. In difficult cases, an additional 3.9-mm trocar (Karl Storz GmbH & Co KG, Tuttlingen, Germany) was used at the right costal margin. In all cases, a retrieval bag (Endo Catch Gold, Covidien, MA) was used.

mbH & Co KG, Tuttlingen, Germany). The 2 small incisions in the posterior vault were closed with resorbable sutures. In difficult cases, an additional 3.9-mm trocar (Karl Storz GmbH & Co KG, Tuttlingen, Germany) was used at the right costal margin. In all cases, a retrieval bag (Endo Catch Gold, Covidien, MA) was used. The 3-trocar NC was performed in the supine position with splayed legs. The first surgeon was standing on the left of the patient, the second surgeon between the legs. Three trocars were used: one umbilical 11-mm and two 3.9-mm trocars (Karl Storz GmbH & Co KG, Tuttlingen, Germany) in the epigastrium and at the right costal margin. A 12-mm Hg capnoperitoneum was established via the first 11-mm trocar, which was inserted via a minilaparotomy at the umbilicus. Dissection technique of the gallbladder was the same in both groups. After preparation and visual confirmation of the cystic duct and the cystic artery, they were transected after the application of clips at the respective distal and proximal end. This was followed by retrograde dissection of the gallbladder using a cautery hook. The gallbladder was extricated in a retrieval bag (ExBag, Medi-Globe GmbH, Achenmuehle, Germany) through the umbilical trocar incision. For clipping (Lapro-clip, Covidien, MA) the cystic duct and the cystic artery via the umbilical trocar and for retrieving the gallbladder, a 3.3-mm optic (30 degrees, 25 cm long, Karl Storz GmbH & Co KG, Tuttlingen, Germany) was used via the epigastric trocar. In cases with large or multiple concretions, the incision including skin and fascia was extended accordingly. The procedure was completed by obligatory closure of the fascia and intracutaneous resorbable sutures. All trocars in both groups were reusable.

& Co KG, Tuttlingen, Germany) was used via the epigastric trocar. In cases with large or multiple concretions, the incision including skin and fascia was extended accordingly. The procedure was completed by obligatory closure of the fascia and intracutaneous resorbable sutures. All trocars in both groups were reusable. Both techniques allow an intraoperative cholangiogram. It could be done via the cystic duct using a percutaneously inserted catheter with the injection of an contrast agent. The first surgeon was the same in all procedures of both groups (DB) to eliminate surgeon-related interindividual effects. Outcome Measures The primary outcome measure was cumulative intensity of postoperative pain in motion, 6 hours after surgery, on POD 1 (2 measures: in the morning and in the evening), and on POD 2 (in the morning), using the numeric rating scale (NRS-11) ranging from 0 (no pain) to 10 (worst imaginable pain).15

The first surgeon was the same in all procedures of both groups (DB) to eliminate surgeon-related interindividual effects. Outcome Measures The primary outcome measure was cumulative intensity of postoperative pain in motion, 6 hours after surgery, on POD 1 (2 measures: in the morning and in the evening), and on POD 2 (in the morning), using the numeric rating scale (NRS-11) ranging from 0 (no pain) to 10 (worst imaginable pain).15 Secondary outcome measures were as follows the aesthetic aspects of the abdominal wall incisions on POD 10 on a 1 (complete satisfaction) to 5 (complete dissatisfaction) scale from patient's and surgeon's point of view, intraoperative complications, conversion rate to classical laparoscopic or open technique, procedural time, assessment of the surgical handling (instrument handling, camera handling, preparation, and gallbladder extraction) by the first and second surgeons on an ordinal scale (from 1 showing “without any problems” to 5 showing “very difficult”), cumulative intensity of postoperative pain in motion from the day of surgery to POD 10 (assessed 6 hours after surgery and at all other days in the morning and in the evening: 21 measurements, using the NRS-11 ranging from 0 to 10), cumulative use of peripheral (paracetamol) and centrally acting analgesics (piritramide) during the first 10 days, postoperative complications, need of reoperation, return to everyday life, and the quality of life on POD 10, assessed using the Gastrointestinal Quality of Life Index (GIQLI; higher score indicates better quality of life) as developed by Eypasch et al.16

trally acting analgesics (piritramide) during the first 10 days, postoperative complications, need of reoperation, return to everyday life, and the quality of life on POD 10, assessed using the Gastrointestinal Quality of Life Index (GIQLI; higher score indicates better quality of life) as developed by Eypasch et al.16 The Clavien-Dindo classification was used to assess postoperative complications.17,18 Conversion was defined either as performing a laparotomy (conversion to conventional procedure) or as the necessity to apply additional or 5-mm trocars (conversion to traditional laparoscopic surgery). In the TVC group, using a 3-mm trocar at the right costal margin was permitted, but insertion of 2 additional 3-mm trocars was considered a conversion to the 3-trocar NC. Studied Data Preoperative, intraoperative, and early postoperative data of both groups were analyzed. The following parameters were prospectively documented: age, BMI, ASA score, number of gallstones (solitary or multiple), size of the biggest gallstone (mm), previous cholecystitis, pre- and postoperative laboratory values [leukocyte count and C-reactive protein (CRP)], number of percutaneous trocars, histopathologic results, and postoperative duration of hospital stay. We report the primary outcome measure and related secondary outcome measures, and put these results into context. A separate paper concerning the long-term results of this study is in preparation.

Studied Data Preoperative, intraoperative, and early postoperative data of both groups were analyzed. The following parameters were prospectively documented: age, BMI, ASA score, number of gallstones (solitary or multiple), size of the biggest gallstone (mm), previous cholecystitis, pre- and postoperative laboratory values [leukocyte count and C-reactive protein (CRP)], number of percutaneous trocars, histopathologic results, and postoperative duration of hospital stay. We report the primary outcome measure and related secondary outcome measures, and put these results into context. A separate paper concerning the long-term results of this study is in preparation. Calculation of Power and Sample Size For the primary endpoint “postoperative pain,” estimates for standard deviation (SD) and minimal clinically relevant difference were derived from the existing literature. For the sample size calculation, a pain reduction of 1.5 points on the NRS-11 was considered clinically relevant.19 The SD for postoperative pain assessment is known to be about 1.5 points so that the expected difference was about 1 SD. To prove this difference in a 2-sided superiority testing with an alpha error of 0.05 and a type 2 error of 0.20 (power 80%), a number of 17 patients for each group was calculated. Because of an expected dropout rate of 10% and the intended use of nonparametric statistics, 40 patients (2 × 20) were included.

Calculation of Power and Sample Size For the primary endpoint “postoperative pain,” estimates for standard deviation (SD) and minimal clinically relevant difference were derived from the existing literature. For the sample size calculation, a pain reduction of 1.5 points on the NRS-11 was considered clinically relevant.19 The SD for postoperative pain assessment is known to be about 1.5 points so that the expected difference was about 1 SD. To prove this difference in a 2-sided superiority testing with an alpha error of 0.05 and a type 2 error of 0.20 (power 80%), a number of 17 patients for each group was calculated. Because of an expected dropout rate of 10% and the intended use of nonparametric statistics, 40 patients (2 × 20) were included. Statistical Analysis IBM SPSS Statistics 19 (IBM Corp., Armonk, NY) was used for data processing and statistics of all variables. All analyses were by intention to treat. Normally distributed parameters such as age and BMI were analyzed using a 2-tailed t test. Not normally distributed parameters were analyzed using the Mann-Whitney U test. Dichotomous questions and all yes/no variables were analyzed using the Fisher exact test for categorical variables. The χ2 test for the trend was used to analyze all ordinal parameters. P < 0.05 was considered statistically significant. This study is registered in the ClinicalTrials.gov Register, ID: NCT01685775, and in the German Clinical Trials Register, ID: DRKS00000341. The Universal Trial Number was U1111-1114-7386.

Statistical Analysis IBM SPSS Statistics 19 (IBM Corp., Armonk, NY) was used for data processing and statistics of all variables. All analyses were by intention to treat. Normally distributed parameters such as age and BMI were analyzed using a 2-tailed t test. Not normally distributed parameters were analyzed using the Mann-Whitney U test. Dichotomous questions and all yes/no variables were analyzed using the Fisher exact test for categorical variables. The χ2 test for the trend was used to analyze all ordinal parameters. P < 0.05 was considered statistically significant. This study is registered in the ClinicalTrials.gov Register, ID: NCT01685775, and in the German Clinical Trials Register, ID: DRKS00000341. The Universal Trial Number was U1111-1114-7386. RESULTS Between February 2010 and June 2012, 40 patients were recruited and randomized (20 in the TVC group and 20 in the NC group). Figure 1 shows the trial profile. Because no conversions were necessary and all patients could be treated according to the study protocol, no further per-protocol analysis was necessary apart from the intention-to-treat analysis. FIGURE 1. Trial profile.

RESULTS Between February 2010 and June 2012, 40 patients were recruited and randomized (20 in the TVC group and 20 in the NC group). Figure 1 shows the trial profile. Because no conversions were necessary and all patients could be treated according to the study protocol, no further per-protocol analysis was necessary apart from the intention-to-treat analysis. FIGURE 1. Trial profile. Table 1 shows the baseline characteristics of all patients. No statistical differences were observed between the group's demographic characteristics, BMI, ASA scores, number and size of gallstones, previous cholecystitis, and preoperative laboratory values (leukocyte count and CRP). The preoperative gynecological examination was without any pathological finding for all TVC patients and had no influence on further procedures. The gynecological examination of all 20 TVC patients conducted 12 to 14 days postoperatively again showed no pathological findings, especially no wound infections. In all cases, suture material was found in place without irritation. TABLE 1. Baseline Characteristics of All Patients

Table 1 shows the baseline characteristics of all patients. No statistical differences were observed between the group's demographic characteristics, BMI, ASA scores, number and size of gallstones, previous cholecystitis, and preoperative laboratory values (leukocyte count and CRP). The preoperative gynecological examination was without any pathological finding for all TVC patients and had no influence on further procedures. The gynecological examination of all 20 TVC patients conducted 12 to 14 days postoperatively again showed no pathological findings, especially no wound infections. In all cases, suture material was found in place without irritation. TABLE 1. Baseline Characteristics of All Patients Variable TVC Group (n = 20) NC Group (n = 20) Total (n = 40) P Age, yr: mean (standard deviation) 44.8 (15.24) 47.5 (18.82) 46.2 (16.67) 0.615* BMI, kg/m2: mean (standard deviation) 28.1 (4.21) 28.5 (4.46) 28.3 (4.29) 0.780* ASA scores 0.358† 1 2 (10%) 6 (30%) 8 (20%) 2 17 (85%) 12 (60%) 29 (72.5%) 3 1 (5%) 2 (10%) 3 (7.5%) Gallbladder stones 0.235‡ Solitaire 2 (10%) 6 (30%) 8 (20%) Multiple 18 (90%) 14 (70%) 32 (80%) Size of the biggest gallstone, mm: median (Q1–Q3) 10.5 (7.0–17.3) 8.0 (4.0–15.0) 10.0 (5.0–15.0) 0.367§ Previous cholecystitis 3 (15%) 2 (10%) 5 (12.5%) 1.000‡ Preoperative laboratory values Leukocyte count, per nL: median (Q1–Q3) 6.9 (4.6–7.8) 7.1 (5.9–9.1) 6.9 (5.3–8.1) 0.173§ CRP, mg/L: median (Q1–Q3) 0.0 (0.0–5.6) 3.3 (0.0–4.3) 3.0 (0.0–4.4) 0.684§ Executed procedure 1.000‡ TVC 20 (100%) 0 (0%) NC 0 (0%) 20 (100%) *t test. †χ2 test for the trend. ‡ Fisher exact test.

Variable TVC Group (n = 20) NC Group (n = 20) Total (n = 40) P Age, yr: mean (standard deviation) 44.8 (15.24) 47.5 (18.82) 46.2 (16.67) 0.615* BMI, kg/m2: mean (standard deviation) 28.1 (4.21) 28.5 (4.46) 28.3 (4.29) 0.780* ASA scores 0.358† 1 2 (10%) 6 (30%) 8 (20%) 2 17 (85%) 12 (60%) 29 (72.5%) 3 1 (5%) 2 (10%) 3 (7.5%) Gallbladder stones 0.235‡ Solitaire 2 (10%) 6 (30%) 8 (20%) Multiple 18 (90%) 14 (70%) 32 (80%) Size of the biggest gallstone, mm: median (Q1–Q3) 10.5 (7.0–17.3) 8.0 (4.0–15.0) 10.0 (5.0–15.0) 0.367§ Previous cholecystitis 3 (15%) 2 (10%) 5 (12.5%) 1.000‡ Preoperative laboratory values Leukocyte count, per nL: median (Q1–Q3) 6.9 (4.6–7.8) 7.1 (5.9–9.1) 6.9 (5.3–8.1) 0.173§ CRP, mg/L: median (Q1–Q3) 0.0 (0.0–5.6) 3.3 (0.0–4.3) 3.0 (0.0–4.4) 0.684§ Executed procedure 1.000‡ TVC 20 (100%) 0 (0%) NC 0 (0%) 20 (100%) *t test. †χ2 test for the trend. ‡ Fisher exact test. §Mann-Whitney U test.

Variable TVC Group (n = 20) NC Group (n = 20) Total (n = 40) P Age, yr: mean (standard deviation) 44.8 (15.24) 47.5 (18.82) 46.2 (16.67) 0.615* BMI, kg/m2: mean (standard deviation) 28.1 (4.21) 28.5 (4.46) 28.3 (4.29) 0.780* ASA scores 0.358† 1 2 (10%) 6 (30%) 8 (20%) 2 17 (85%) 12 (60%) 29 (72.5%) 3 1 (5%) 2 (10%) 3 (7.5%) Gallbladder stones 0.235‡ Solitaire 2 (10%) 6 (30%) 8 (20%) Multiple 18 (90%) 14 (70%) 32 (80%) Size of the biggest gallstone, mm: median (Q1–Q3) 10.5 (7.0–17.3) 8.0 (4.0–15.0) 10.0 (5.0–15.0) 0.367§ Previous cholecystitis 3 (15%) 2 (10%) 5 (12.5%) 1.000‡ Preoperative laboratory values Leukocyte count, per nL: median (Q1–Q3) 6.9 (4.6–7.8) 7.1 (5.9–9.1) 6.9 (5.3–8.1) 0.173§ CRP, mg/L: median (Q1–Q3) 0.0 (0.0–5.6) 3.3 (0.0–4.3) 3.0 (0.0–4.4) 0.684§ Executed procedure 1.000‡ TVC 20 (100%) 0 (0%) NC 0 (0%) 20 (100%) *t test. †χ2 test for the trend. ‡ Fisher exact test. §Mann-Whitney U test. Table 2 shows the procedural data and outcomes. No significant differences were found for procedural time, frequency of analgesics in the recovery room, pre- and postoperative difference in CRP and leukocyte count, postoperative hospital stay, postoperative complications, and time to everyday life. As expected, the median amount of percutaneous trocars was significantly less in the TVC group than in the NC group (1 vs 3; P < 0.001). No patient suffered from an intraoperative complication, and there was no case with a need for conversion, drain placement, blood transfusion, or revision surgery. There were neither pre- nor intraoperative findings that suggested choledocholithiasis in any of the patients. Therefore, an intraoperative cholangiogram was not deemed necessary in any patient. Mortality was 0 in both groups. In each group, there were 2 postoperative complications. Two wound infections at the umbilical trocar site, which was also used for the retrieval of the gallbladder, occurred in the NC group. One wound had to be opened and needed regular flushing, whereas the other healed primarily under antibiotic treatment. Contrarily, the 2 postoperative complications in the TVC group were biliary pancreatitis, one of them with postoperative cholestasis and cystic duct leakage. In both cases, an endoscopic retrograde cholangiopancreatography was performed and after the according conservative therapy and prolonged hospital stay, both patients were discharged free of symptoms. Preoperatively, even in these 2 cases, there were no clinical, ultrasonographic, or laboratory (liver function test) findings suspect for choledocholithiasis. Thus, independently of the procedural technique, no indication for intraoperative cholangiogram existed. The resulting Clavien-Dindo classifications are listed in Table 3. Compared with the TVC patients without complications, the 2 patients with the biliary pancreatitis had the highest demand for peripheral and centrally acting analgesics (median, 11.8 g vs 4.0 g; 180 mg vs 6.25 mg) and the highest cumulative pain score from the day of surgery until POD 2 and until POD 10 (median, 5 vs 2; 76.5 vs 17).

the TVC patients without complications, the 2 patients with the biliary pancreatitis had the highest demand for peripheral and centrally acting analgesics (median, 11.8 g vs 4.0 g; 180 mg vs 6.25 mg) and the highest cumulative pain score from the day of surgery until POD 2 and until POD 10 (median, 5 vs 2; 76.5 vs 17). They also had the lowest GIQLI values (median, 75.5 vs 124). TABLE 2. Procedural Data and Outcomes Variable TVC Group (n = 20) NC Group (n = 20) Total (n = 40) P Procedural time, min: median (Q1–Q3) 50.0 (42.0–66.0) 54.5 (46.0–62.0) 52.0 (44.0–62.0) 0.675* No.

the TVC patients without complications, the 2 patients with the biliary pancreatitis had the highest demand for peripheral and centrally acting analgesics (median, 11.8 g vs 4.0 g; 180 mg vs 6.25 mg) and the highest cumulative pain score from the day of surgery until POD 2 and until POD 10 (median, 5 vs 2; 76.5 vs 17). They also had the lowest GIQLI values (median, 75.5 vs 124). TABLE 2. Procedural Data and Outcomes Variable TVC Group (n = 20) NC Group (n = 20) Total (n = 40) P Procedural time, min: median (Q1–Q3) 50.0 (42.0–66.0) 54.5 (46.0–62.0) 52.0 (44.0–62.0) 0.675* No. of percutaneous trocars: median (Q1–Q3) 1 (1–1) 3 (3–3) 2.5 (1–3) <0.001* Intraoperative complication 0 (0%) 0 (0%) 0 (0%) 1.000† Conversion 0 (0%) 0 (0%) 0 (0%) 1.000† Surgical handling, assessment by the first surgeon, 1/2/3/4/5‡ Instrument handling 2/12/6/0/0 4/14/2/0/0 6/26/8/0/0 0.112§ Camera handling 2/14/3/1/0 0/14/6/0/0 2/28/9/1/0 0.411§ Preparation 2/14/3/1/0 3/9/8/0/0 5/23/11/1/0 0.645§ Gallbladder extraction 17/3/0/0/0 3/7/8/2/0 20/10/8/2/0 <0.001§ Surgical handling, assessment by the second surgeon, 1/2/3/4/5‡ Instrument handling 5/6/7/2/0 6/14/0/0/0 11/20/7/2/0 0.020§ Camera handling 4/11/4/1/0 4/9/7/0/0 8/20/11/1/0 0.835§ Preparation 11/5/3/1/0 4/13/3/0/0 15/18/6/1/0 0.311§ Gallbladder extraction 9/10/1/0/0 3/11/4/2/0 12/21/5/2/0 0.010§ Drainage 0 (0%) 0 (0%) 0 (0%) 1.000† Blood transfusion 0 (0%) 0 (0%) 0 (0%) 1.000† Cumulative pain from the day of surgery to day 2 in the morning, NRS-11: median (Q1–Q3) 8.0 (4–16)¶ 14.0 (8–19)¶ 12.0 (5.8–17.5)|| 0.043* Cumulative pain from the day of surgery to day 10, NRS-11: median (Q1–Q3) 22.0 (11.5–37.5)¶ 41.0 (26.0–66.0)¶ 28.5 (16.3–49.0)|| 0.010* Frequency of analgesics in the recovery room 6 (30%) 7 (35%) 13 (32.5%) 1.000† Cumulative paracetamol, g, from the day of surgery to day 10: median (Q1–Q3) 4.5 (3.0–8.5)¶ 8.75 (5.0–14.5)¶ 6.5 (3.5–10.0)|| 0.019* Cumulative piritramide, mg, from the day of surgery to day 10: median (Q1–Q3) 7.5 (0.0–22.5)¶ 0.0 (0.0–17.5)¶ 6.3 (0.0–21.4)|| 0.244* Pre- and postoperative difference in CRP, mg/L: median (Q1–Q3) 12.6 (3.6–26.1) 10.3 (4.9–22.6) 12.3 (4.0–24.79) 0.574* Pre- and postoperative difference in leukocyte count, per nL: median (Q1–Q3) 0.5 (−0.7–1.3) 0.1 (−0.7–0.6) 0.1 (−0.7–1.0) 0.496* Postoperative complication 2 (10%) 2 (10%) 4 (10%) 1.000† Revision surgery 0 (0%) 0 (0%) 0 (0%) 1.000† Postoperative hospital stay, d: median (Q1–Q3) 2.0 (2–2) 2.0 (2–2) 2.0 (2.0–2.0) 0.892* Time to everyday life, d: median (Q1–Q3) 5.5 (3.5–12.5) 6.5 (6.0–11.0) 6.0 (4.3–12.8) 0.328* Assessment of aesthetic on postoperative day 10, 1/2/3/4/5** Patient 20/0/0/0/0 6/11/3/0/0 26/11/3/0/0 <0.001§ Investigator 20/0/0/0/0 0/16/4/0/0 20/16/4/0/0 <0.001§ Gastrointestinal Quality of Life Index

) 2.0 (2.0–2.0) 0.892* Time to everyday life, d: median (Q1–Q3) 5.5 (3.5–12.5) 6.5 (6.0–11.0) 6.0 (4.3–12.8) 0.328* Assessment of aesthetic on postoperative day 10, 1/2/3/4/5** Patient 20/0/0/0/0 6/11/3/0/0 26/11/3/0/0 <0.001§ Investigator 20/0/0/0/0 0/16/4/0/0 20/16/4/0/0 <0.001§ Gastrointestinal Quality of Life Index on postoperative day 10: median (Q1–Q3) 124 (104.5–132.0) 107 (89.0–114.0) 115.5 (99.3–125.5) 0.028* Mortality 0 (0%) 0 (0%) 0 (0%) 1.000† *Mann-Whitney U test. †Fisher exact test. ‡Response options: 1, without any problems; 2, easy; 3, moderate; 4, slightly difficult; 5, very difficult. §χ2 test for the trend. ¶n = 19. ∥n = 38. **Response options: 1, very satisfied; 2, moderately satisfied; 3, equally satisfied and dissatisfied; 4, moderately dissatisfied; 5, very dissatisfied. TABLE 3. Clavien-Dindo Classification of Postoperative Complications TVC Group (n = 20) LC Group (n = 20) Total (n = 40) P 0.852* No complication 18 (90%) 18 (90%) 36 (90%) Grade I 0 (0%) 0 (0%) 0 (0%) Grade II 0 (0%) 1 (5%) 1 (2.5%) Grade III 2 (10%) 1 (5%) 3 (7.5%) Grade IV 0 (0%) 0 (0%) 0 (0%) Grade V 0 (0%) 0 (0%) 0 (0%) *χ2 test for the trend. Grade I: no intervention necessary. Grade II: requiring pharmacological treatment. Grade III: requiring surgical, endoscopic, or radiological intervention. Grade IV: life-threatening complication requiring intensive care/intensive care unit management. Grade V: death of a patient. One patient of each group did not fill out the pain and analgesic diary prospectively, so they had to drop out of the analysis of postoperative pain.

Grade III: requiring surgical, endoscopic, or radiological intervention. Grade IV: life-threatening complication requiring intensive care/intensive care unit management. Grade V: death of a patient. One patient of each group did not fill out the pain and analgesic diary prospectively, so they had to drop out of the analysis of postoperative pain. The primary outcome measure, the cumulative intensity of postoperative pain in motion for the first 48 hours postoperatively was significantly lower in the TVC group (Table 2). Figure 2 shows boxplots for the 4 single NRS-11 scores of both groups compared with the respective quarter overall median. FIGURE 2. Boxplots for the primary outcome measure (single NRS-11 scores of both groups compared with the respective quarter overall median).

The primary outcome measure, the cumulative intensity of postoperative pain in motion for the first 48 hours postoperatively was significantly lower in the TVC group (Table 2). Figure 2 shows boxplots for the 4 single NRS-11 scores of both groups compared with the respective quarter overall median. FIGURE 2. Boxplots for the primary outcome measure (single NRS-11 scores of both groups compared with the respective quarter overall median). Furthermore, the cumulative intensity of pain from the day of surgery to POD 10 was significantly lower for TVC patients, although they needed significantly less analgesics until POD 10. As mentioned earlier, the trend for a higher consumption of centrally acting analgesics can mostly be attributed to the 2 TVC patients with postoperative complications. After exclusion of the patients with postoperative complications of both groups, piritramide consumption in the TVC group even showed a trend to be lower (median/Q1–Q3, 5.0/0–16.5 mg vs 7.5/0–25.0 mg; P = 0.740). In addition, after exclusion of the patients with postoperative complications, the primary outcome measure, the second pain evaluation and the postoperative need for peripheral analgesics, resulted in even more pronounced significance toward TVC (NRS-11, 7.0/3.5–14.0 vs 14.0/7.5–21.5; P = 0.017; NRS-11, 17.0/5.5–30.5 vs 41.0/25.5–55.5; P = 0.002; 4.0/2.5–6.5 g vs 8.5/4.5–15.0 g; P = 0.013). In this analysis, there was also a greater difference in time to everyday life, which, however, was not quite significant (5.0/3.0–10.0 days vs 7.0/6.0–13.5 days; P = 0.075).

RS-11, 7.0/3.5–14.0 vs 14.0/7.5–21.5; P = 0.017; NRS-11, 17.0/5.5–30.5 vs 41.0/25.5–55.5; P = 0.002; 4.0/2.5–6.5 g vs 8.5/4.5–15.0 g; P = 0.013). In this analysis, there was also a greater difference in time to everyday life, which, however, was not quite significant (5.0/3.0–10.0 days vs 7.0/6.0–13.5 days; P = 0.075). Gallbladder extraction in the TVC group was rated to be significantly easier by both surgeons, whereas handling of instruments was considered significantly more difficult by the 2 surgeons for the TVC group. Camera handling and preparation were not rated different between both groups. Evaluation of the GIQLI by the patient and the aesthetic result on POD 10 by the patient and the investigator was significantly better after TVC. There was no readmission of any patient in either group during follow-up. DISCUSSION This trial was designed to compare the intensity of postoperative pain after 2 different techniques of cholecystectomy in patients with symptomatic cholecystolithiasis. Results are significantly reduced postoperative pain in the first 2 days and in the first 10 days after surgery by TVC compared with NC. Need of peripheral analgesics was significantly less in the TVC group. TVC patients were significantly more satisfied with the aesthetic result and had a significantly better GIQLI.

. Results are significantly reduced postoperative pain in the first 2 days and in the first 10 days after surgery by TVC compared with NC. Need of peripheral analgesics was significantly less in the TVC group. TVC patients were significantly more satisfied with the aesthetic result and had a significantly better GIQLI. For many years now, a patient-directed aim of operative procedures is the reduction of access-related trauma. Hereby, postoperative pain can be reduced and access-related complications such as wound infections, cutaneous scars, adhesions, and incisional hernias avoided. This leads to a quicker recovery, shorter hospital stay, a better aesthetic result, and an increased postoperative quality of life, as proven for some procedures after the introduction of laparoscopic surgery.20–22 The disadvantages of conventional surgery persist in laparoscopic procedures that require a retrieval incision for the specimen. Therefore, the concept of access to the abdominal cavity through already existing, natural routes, namely NOTES, evolved. After this concept, gallbladders with large concretions were retrieved through a posterior colpotomy during LC in female patients as early as 1993.23 Even for laparoscopic splenectomy, transvaginal retrieval was described early.24 In 2007, several groups published different techniques of TVC for the first time.9,25–27 They used either a flexible endoscope or, like Zornig et al,9 a regular laparoscopic optic. For retraction of the gallbladder or for preparation, one or more percutaneous trocars were used. For example, in the hybrid—NOS—technique of Zornig et al,9 a transumbilical 5-mm trocar, a transvaginal 10-mm –trocar, and a transvaginal rigid 5-mm grasping forceps are used. Meanwhile, pure NOTES techniques exist, but in clinical routine, the hybrid technique is widely used.8 After feasibility and safety of the TVC using rigid instruments and laparoscopes was shown in case series,28–33 it was confirmed in nonrandomized studies comparing TVC and standard LC.34,35 More nonrandomized studies even found advantages for TVC as compared with LC. In a retrospective case-controlled study for TVC, as described by Zornig et al,9 compared with LC in a 3-trocar technique in a total of 93 patients, Hensel et al11 found nausea or vomiting, pain, use of analgesics, and hospital stay significantly reduced.

mized studies even found advantages for TVC as compared with LC. In a retrospective case-controlled study for TVC, as described by Zornig et al,9 compared with LC in a 3-trocar technique in a total of 93 patients, Hensel et al11 found nausea or vomiting, pain, use of analgesics, and hospital stay significantly reduced. Also, Kilian et al10 found significantly less postoperative pain and shorter hospital stay for 15 TVC patients (Zornig technique) than for 20 LC patients in a 3-armed nonrandomized study. Postoperative pain values for 14 TVC patients were significantly less on POD 1 and 3 than for 22 single-incision cholecystectomy patients and 11 LC patients in the 3-armed study of Solomon et al.36 Borchert et al37 found significantly less pain and consumption of analgesics on POD 3 in their analysis of 77 TVC and 46 LC patients. Santos et al12 compared 7 TVC patients with 7 standard LC patients using a flexible endoscope. Despite the small sample size, they also found significantly less pain on the day of surgery and on POD 1 and significantly less consumption of centrally acting analgesics in the recovery room. The same results were found in our cohort analysis comparing 50 TVC patients (Zornig technique) and 50 LC patients.13 There was significantly less pain on the first 2 days after surgery, and the use of analgesics in the recovery room was significantly less frequent. All cited studies have in common that TVC was performed as a hybrid procedure. However, until now only one prospective randomized clinical trial dealing with the new techniques of cholecystectomy has been published.38 It is a 3-arm pilot study comparing hybrid NOTES transvaginal, pure transumbilical, and conventional LC. No differences in the complication rate, length of hospital stay, and time off from work were found, but the estimated sample size was underpowered, making interpretation of the results limited. Because of the lack of an adequately powered trial, we designed this first randomized controlled study for the comparison of TVC and NC.

ces in the complication rate, length of hospital stay, and time off from work were found, but the estimated sample size was underpowered, making interpretation of the results limited. Because of the lack of an adequately powered trial, we designed this first randomized controlled study for the comparison of TVC and NC. In our clinic, we began performing TVC in the year 2008 and that we did it in the manner as described by Zornig et al.9 Analysis of the first 50 TVC patients was followed by a cohort analysis of these patients with 50 LC patients, and it found significantly less postoperative pain despite lower amount of analgesics used.13 We then planned this prospective, randomized clinical trial to confirm the results. Because of the data that show advantages for both needlescopic and 3-trocar techniques concerning postoperative pain and postoperative amount of analgesics,5,6,39 we voted for the NC technique as a control group. Despite a higher conversion rate, this technique had better aesthetic results than traditional LC.4 Thus, the advantages we found for TVC compared with NC are even more pronounced than a comparison with traditional LC would have shown.

rative amount of analgesics,5,6,39 we voted for the NC technique as a control group. Despite a higher conversion rate, this technique had better aesthetic results than traditional LC.4 Thus, the advantages we found for TVC compared with NC are even more pronounced than a comparison with traditional LC would have shown. Similarly to the cited nonrandomized trials, the rate of postoperative complications in our trial, namely 2 cases of postoperative biliary pancreatitis after TVC and 2 cases of wound infections after NC, did not account for a significant difference. However, the existing data confirm the retrieval incision as a source for wound infections and the majority of pain.40 Because a retrieval incision in the abdominal wall is not required in TVC, we would expect a significant difference in the rate of wound infections in a study with a higher sample size. On the one hand, our study was not designed and sized for this parameter as a primary outcome, so further investigation is needed to confirm this trend. On the other hand, the primary outcome parameter “reduction of postoperative pain” by avoiding specimen retrieval through the abdominal wall with the TVC technique is markedly confirmed in our trial. For evaluation of a theoretically lower rate of port-site hernias after TVC, the sample size of our study is too small as well and a follow-up of several years is required. In our study there are 2 cases of postoperative biliary pancreatitis in the TVC group. Whether this is a technique-specific complication or a coincidence in a study collective cannot be determined because of the lack of significance. An accumulation of this complication after TVC has not been described in the literature. In our own evaluation of the first 50 TVC patients compared with traditional LC patients, that complication did not occur.13 In contrast to our cohort analysis, no in pre- and postoperative difference in CRP and in postoperative hospital stay between the 2 groups was found in our randomized trial. However, the surgical technique for the control groups differs, which limits comparability between the 2 studies. The missing difference might be attributed to the reduced invasiveness of NC. Interestingly, there is a pronounced difference in mean procedural time. Although procedural time was 77.8 minutes in our cohort analysis, it was 53.6 minutes in this trial.

ol groups differs, which limits comparability between the 2 studies. The missing difference might be attributed to the reduced invasiveness of NC. Interestingly, there is a pronounced difference in mean procedural time. Although procedural time was 77.8 minutes in our cohort analysis, it was 53.6 minutes in this trial. This impressively reflects the procedural learning curve, as the cohort analysis consisted of the first 50 TVC patients who were observed in our clinic. CONCLUSIONS In a randomized clinical trial, needlescopic 3-trocar cholecystectomy was compared to transvaginal/transumbilical hybrid-NOTES-technique for symptomatic cholecystolithiasis. Although comparable in terms of safety, we found in the NOTES-group significantly less pain despite less use analgesics, increased satisfaction with the aesthetic result, and improved postoperative quality of life in the short term. ACKNOWLEDGMENT The authors thank Dr S. Sauerland of the Institute for Research in Operative Medicine, University of Witten/Herdecke, Cologne, Germany, for assistance with the statistical analysis of this study. Disclosure: Supported in part by the German Ministry of Research and Education (CHIR-Net Grant, BMBF No. 01-GH-0605). The authors declare no conflicts of interest.

Road traffic crashes claimed the lives of 1.3 million people globally, and accounted for 34% of all years lived with disability (YLDs) attributed to injury, in 2010.1,2 Over the last 40 years or so, there has been a substantial reduction in road traffic fatality rates in high-income countries,3 including dramatic improvements in Australia4 attributed to legislation (eg, mandatory seat belt legislation, compulsory motorcycle and bicycle helmets, reduced speed limits, and random breath testing for alcohol) and widespread public health campaigns. Primary prevention and efforts to improve care of the injured are important for reducing injury burden,5 but the impact of systems-based approaches to trauma care on burden and cost of road traffic injury has not been evaluated. Traumatic injury is described as a time-sensitive disease, with rapid provision of high-quality care critical to prevent death and disability.6 People involved in road traffic crashes often sustain multiple injuries of higher complexity, commonly in rural or remote locations, creating substantial challenges for organization of emergency and health services.7 Regionalized trauma care, where seriously injured patients are preferentially transported to designated trauma centers, is believed to increase the likelihood of survival.8–10 These systems ideally include primary prevention, coordination of prehospital and hospital care, and provision of rehabilitation and postdischarge care to reduce preventable prehospital and in-hospital deaths and minimize long-term complications and disability.6,11

s believed to increase the likelihood of survival.8–10 These systems ideally include primary prevention, coordination of prehospital and hospital care, and provision of rehabilitation and postdischarge care to reduce preventable prehospital and in-hospital deaths and minimize long-term complications and disability.6,11 Evidence supports the view that regionalized trauma care reduces mortality.7–10 However, most studies have focused only on cases where hospitalization occurred. The exception is Nathens et al,7 who reported reduced risk of 30-day mortality after motor vehicle crashes in the United States using the fatality reporting system for road trauma. In contrast to mortality outcomes, the evidence for an effect on disability is sparse. Improved survival rates could result in higher or lower rates of severity of disability in survivors. Two prospective studies have shown improved functional outcomes for patients managed at specialized trauma centers.12,13 A comprehensive investigation of the impact of regionalization of care should ideally include prehospital deaths, in-hospital mortality, and long-term disability.

severity of disability in survivors. Two prospective studies have shown improved functional outcomes for patients managed at specialized trauma centers.12,13 A comprehensive investigation of the impact of regionalization of care should ideally include prehospital deaths, in-hospital mortality, and long-term disability. The state of Victoria, Australia, operated an ad hoc system for delivering trauma care throughout the 20th century, and the annual decline in road traffic fatality rates slowed to 1% per annum in the decade to 2002 compared with 5.3% per annum in the previous decade.14 In the late 1990s, growing evidence that a high percentage of road trauma deaths could be prevented by improved care triggered the reorganization of trauma care in the state.15 The change to a regionalized and integrated trauma system was funded by the State Government Department of Health and the state's third party insurer for road trauma, the Transport Accident Commission, with the aims of reducing injury-related mortality and morbidity. The aim of this study was to describe the burden of road transport–related serious injury in Victoria, Australia, over a 10-year period, after the introduction of a regionalized, inclusive trauma system.

or road trauma, the Transport Accident Commission, with the aims of reducing injury-related mortality and morbidity. The aim of this study was to describe the burden of road transport–related serious injury in Victoria, Australia, over a 10-year period, after the introduction of a regionalized, inclusive trauma system. METHODS Setting The state of Victoria, Australia (population 5.6 million), implemented the Victorian State Trauma System between 2000 and 2003.8 The Victorian State Trauma System encompasses all 138 trauma-receiving hospitals in the state, designated according to capability. Three hospitals (2 adult, 1 pediatric) were designated as major trauma services (level 1 trauma center equivalent). A single ambulance service provides road and air (fixed wing and helicopter) transport of patients. The Victorian State Trauma System meets the criteria for inclusive and regionalized trauma care including designation of a small number of major trauma services, prehospital triage guidelines allowing bypass of smaller centers in favor of direct transport to the specialist major trauma service hospitals, agreed interhospital transfer guidelines, quality assurance programs, a statewide monitoring system, and a governance system capable of impacting change.6,11

r trauma services, prehospital triage guidelines allowing bypass of smaller centers in favor of direct transport to the specialist major trauma service hospitals, agreed interhospital transfer guidelines, quality assurance programs, a statewide monitoring system, and a governance system capable of impacting change.6,11 Data Sources Victorian State Trauma Registry The Victorian State Trauma Registry is a statewide, population-based register of all hospitalized major trauma cases in Victoria.16 Data collection commenced in July 2001. A case is included if any of the following criteria are met: (1) death due to injury; (2) an injury severity score (ISS) more than 12 as determined by the abbreviated injury scale (AIS) (2005 version 2008 update); (3) admission to an intensive care unit (ICU) more than 24 hours and requiring mechanical ventilation for at least part of their ICU stay; and (4) urgent surgery. Isolated hip fracture cases are excluded. The registry collects data from the prehospital and acute care hospital phases, and collects longer-term functional and health-related quality-of-life outcomes for all survivors to hospital discharge via telephone interview at 6, 12, and 24 months after injury.16,17

nt surgery. Isolated hip fracture cases are excluded. The registry collects data from the prehospital and acute care hospital phases, and collects longer-term functional and health-related quality-of-life outcomes for all survivors to hospital discharge via telephone interview at 6, 12, and 24 months after injury.16,17 National Coroners' Information System The National Coroners' Information System is an Internet-based data storage and retrieval system for Australian coronial cases (http://www.ncis.org.au/) and includes every death reported to the Victorian coroner since 2000. In Victoria, legislation requires a death that seems to have been unexpected, unnatural, or violent or to have resulted, directly or indirectly, from an accident or injury to be reported to the coroner. To classify cases as an injury-related death, all cases with a cause of death as “death due to external causes,” “still enquiring,” and “unlikely to be known” are reviewed, and deaths involving (1) airway obstruction by a foreign body; (2) asbestosis; (3) poisoning and drug/alcohol overdose; (4) medical/surgical complications; and (5) other nontraumatic incidents not excluded by the filters (eg, malignancy) are excluded.

causes,” “still enquiring,” and “unlikely to be known” are reviewed, and deaths involving (1) airway obstruction by a foreign body; (2) asbestosis; (3) poisoning and drug/alcohol overdose; (4) medical/surgical complications; and (5) other nontraumatic incidents not excluded by the filters (eg, malignancy) are excluded. Ethics Statement The Victorian State Trauma Registry uses an opt-off consent process where all eligible cases are included on the registry and patients (or their next of kin) are provided with a letter and a brochure stating the aims of the registry and data collected.16–18 The brochure provides the details of how to opt-off and the opt-off rate is less than 1%. An opt-off consent is used because of the impracticability of informed consent, and the potential for selection bias, in the registry setting.19 Where patients are followed up after hospital discharge by telephone interview, verbal consent to complete the interview is obtained.17 Access to the National Coroners' Information System data is part of the registry protocol and approved by the custodians of the coroners' data. The Victorian State Trauma Registry protocol, including the described consent process, has been approved by the Human Research Ethics Committee at each participating hospital and the Monash University Human Research Ethics Committee.

is part of the registry protocol and approved by the custodians of the coroners' data. The Victorian State Trauma Registry protocol, including the described consent process, has been approved by the Human Research Ethics Committee at each participating hospital and the Monash University Human Research Ethics Committee. Participants All road transport–related major trauma cases and deaths from July 1, 2001, to June 30, 2011, were extracted from the Victorian State Trauma Registry and the National Coroners' Information System. Road transport cases included motor vehicle, motorcycle, pedestrians, pedal cyclists, and other road transport (eg, heavy vehicle) that occurred on a public road, street, or highway. For the purposes of this study, major trauma was defined as an AIS 2008 version ISS more than 12. Injury diagnoses coded before introduction of the AIS 2008 were mapped from the AIS 1990, 1998 version to the AIS 2008, using a validated map.20 All prehospital and in-hospital deaths were included.

ad, street, or highway. For the purposes of this study, major trauma was defined as an AIS 2008 version ISS more than 12. Injury diagnoses coded before introduction of the AIS 2008 were mapped from the AIS 1990, 1998 version to the AIS 2008, using a validated map.20 All prehospital and in-hospital deaths were included. Procedures and Data Analysis Incidence of Trauma-Related Mortality and Major Trauma Population-based incidence rates, and 95% confidence intervals (95% CIs), were calculated for each financial year based on the total population on June 30 for the years 2002–2011. Poisson regression was used to determine whether the incidence rate increased or decreased over the 10-year period. A check for potential overdispersion of the data (variance greater than the mean) was performed to ensure that the assumptions of a Poisson distribution were met. Models were fitted for all road transport–related deaths and for major trauma, and the incidence rate ratio and 95% CI reported.

sed over the 10-year period. A check for potential overdispersion of the data (variance greater than the mean) was performed to ensure that the assumptions of a Poisson distribution were met. Models were fitted for all road transport–related deaths and for major trauma, and the incidence rate ratio and 95% CI reported. Risk-Adjusted In-hospital Mortality Descriptive statistics were used to provide an overview of the pattern of hospitalized road transport–related major trauma patients captured by the Victorian State Trauma Registry by year. A logistic regression model was fitted with in-hospital mortality as the outcome and year as the covariate of interest. The model was adjusted for known predictors of mortality in our population21 and identified changes in case-mix over time. The covariates used were age, natural logarithm of the ISS, head injury severity according to the AIS severity score (0–2, 3–4, 5–6), comorbid status as measured by the Charlson Comorbidity Index weighting, and road user group, consistent with trauma mortality risk prediction models.22,23 Adjusted odds ratios and 95% CIs for each financial year, relative to the 2001–2002 year, were calculated.

rity according to the AIS severity score (0–2, 3–4, 5–6), comorbid status as measured by the Charlson Comorbidity Index weighting, and road user group, consistent with trauma mortality risk prediction models.22,23 Adjusted odds ratios and 95% CIs for each financial year, relative to the 2001–2002 year, were calculated. Disability-Adjusted Life Years The disability-adjusted life year (DALY) is a common metric for measuring disease burden or “health loss.”24 The DALY combines years of life lost (YLLs) and YLDs to generate DALYs for diseases or conditions.25 The YLL component is calculated by multiplying the number of deaths by the standard life expectancy at age of death in years. Life expectancy was obtained from the 2008–2010 Australian standard life table. For each year, the number of deaths occurring in each age group and sex was used to establish the YLLs, thereby accounting for differences in age and sex profiles over time. Patient identifying information was used to cross-check between the National Coroners' Information System and the Victorian State Trauma Registry data to ensure that deaths were only counted once.

ach age group and sex was used to establish the YLLs, thereby accounting for differences in age and sex profiles over time. Patient identifying information was used to cross-check between the National Coroners' Information System and the Victorian State Trauma Registry data to ensure that deaths were only counted once. The YLD component is calculated as the product of the number of incident cases in the time period multiplied by the average duration of the disease (years expected to life in the disabled state) and a weight factor that reflects the severity of the disease on a scale from 0 (perfect health) to 1 (dead). Previously published disability weights, including weights from the 2010 Global Burden of Disease Study,24 have been based on the disability experienced after sustaining a single injury. As the majority of major trauma patients, particularly those injured in road traffic crashes, sustain multiple injuries, weights were developed specifically for this project from the long-term outcome data collected by the Victorian State Trauma Registry. The UK Burden of Injury project approach was followed,26 except that the injuries sustained by the patients were mapped to 7 injury groups, based on the AIS classification rather than the 10th revision of the International Classification of Diseases, and multiple injuries were accounted for in the groupings.

Registry. The UK Burden of Injury project approach was followed,26 except that the injuries sustained by the patients were mapped to 7 injury groups, based on the AIS classification rather than the 10th revision of the International Classification of Diseases, and multiple injuries were accounted for in the groupings. Disability weights were calculated using the EQ-5D. This is a generic measure of health status consisting of 5 questions, which ask about mobility, self-care, usual activities, pain or discomfort, and anxiety or depression.27 Disability weights were calculated from the EQ-5D responses of 3505 adult Victorian State Trauma Registry cases (road transport and other cause) with an ISS more than 12, who were eligible for 6, 12, and 24-month follow-up between January 2009 and December 2010. This time frame was chosen as it reflects the timing of the introduction of the EQ-5D to the registry follow-up protocol. Responses to the 5 items were converted into a summary social preference index score ranging from less than 0 (representing a health state worse than death) to 1 (best health) using UK tariffs.28 Proxy responses were substituted where patient responses were not available.29

the EQ-5D to the registry follow-up protocol. Responses to the 5 items were converted into a summary social preference index score ranging from less than 0 (representing a health state worse than death) to 1 (best health) using UK tariffs.28 Proxy responses were substituted where patient responses were not available.29 Disability weights were calculated by subtracting the patient (or proxy) score from the relevant age and sex population–matched norm. The average weight at each follow-up time point, in each injury group, was calculated, along with the standard error of the weight, to provide a measure of the level of precision of the estimated disability weight. Short-term weights were the time-weighted mean disability weight for the first 24 months postinjury, whereas the mean 24-month weight was considered the long-term weight for the injury group. All patients (or their proxy respondent) were asked to rate the patient's level of disability before injury and at follow-up on a 5-point scale from none to severe disability.30 Residual disability at 24 months was confirmed if the level of disability reported at 24 months was greater than preinjury disability, and this was considered permanent for the purposes of calculating YLDs. The proportion of patients with residual disability at 24 months, and the corresponding 95% CI, was reported.

.30 Residual disability at 24 months was confirmed if the level of disability reported at 24 months was greater than preinjury disability, and this was considered permanent for the purposes of calculating YLDs. The proportion of patients with residual disability at 24 months, and the corresponding 95% CI, was reported. The number of cases in each injury group and age group, separated by sex, was multiplied by the relevant disability weight and duration of disability to calculate YLDs. These calculations were performed separately for each year, thereby accounting for differences in the age, sex, and injury case-mix occurring in each financial year. Total DALYs were calculated by adding the YLDs and YLLs in each year. Consistent with the 2010 Global Burden of Disease study, age discounting was not used.31 Economic discounting at 3% was used, which is consistent with World Health Organization recommendations for burden of disease studies.32

in each financial year. Total DALYs were calculated by adding the YLDs and YLLs in each year. Consistent with the 2010 Global Burden of Disease study, age discounting was not used.31 Economic discounting at 3% was used, which is consistent with World Health Organization recommendations for burden of disease studies.32 Costs Establishing an economic cost to the burden of road-related injury requires a dollar cost per DALY. There is no consensus, with the value placed on a healthy year of life varying depending on the circumstances. For this study, 2 measures of cost per DALY were applied. The first was A$50,000 per DALY, which reflects Australian gross domestic product (GDP) per capita and empirical data about what is considered acceptable value for money in Australia.33 The second was A$151,000 per DALY, which is the value of a statistical life year (VSLY) in 2007 recommended by the Australian Government Department of Finance and Deregulation after a review of approaches to cost-benefit analysis.34 The VSLY is an estimate of the social willingness to pay to reduce premature death, which is considered the appropriate way to estimate the value of reduced risk of physical harm.

007 recommended by the Australian Government Department of Finance and Deregulation after a review of approaches to cost-benefit analysis.34 The VSLY is an estimate of the social willingness to pay to reduce premature death, which is considered the appropriate way to estimate the value of reduced risk of physical harm. RESULTS Incidence of Road Transport–Related Mortality and Hospitalized Major Trauma There were 7828 hospitalized road transport–related major trauma (ISS >12) cases in Victoria over the 10 years; 4562 motor vehicle occupants, 1426 motorcyclists, 1199 pedestrians, and 641 other road users. The incidence of hospitalized road transport–related major trauma increased over the decade (incidence rate ratio 1.03, 95% CI: 1.02–1.04; P < 0.001) (Fig. 1). FIGURE 1. Incidence of transport-related deaths and hospitalized major trauma in Victoria (July 2001 to June 2011). There were 3436 road transport–related deaths in Victoria over the 10-year period; 2347 motor vehicle occupants, 528 pedestrians, 457 motorcyclists, and 104 other road users. There was a significant reduction in the population incidence of all road transport–related deaths (incidence rate ratio 0.95, 95% CI: 0.94–0.96; P < 0.001) (Fig. 1).

transport–related deaths in Victoria over the 10-year period; 2347 motor vehicle occupants, 528 pedestrians, 457 motorcyclists, and 104 other road users. There was a significant reduction in the population incidence of all road transport–related deaths (incidence rate ratio 0.95, 95% CI: 0.94–0.96; P < 0.001) (Fig. 1). Risk-Adjusted In-hospital Mortality The pattern of hospitalized road transport–related major trauma changed over time (Table 1). The mean age of injured patients increased, as did the proportion of cases with substantial comorbidities. There was no clear change in the ISS of patients, but the proportion with an intracranial injury (AIS head Injury Severity Score ≤2) decreased over time (Table 1). Direct transport from the scene of injury to a major trauma service increased from 55% of cases in 2001–2002 to 77% of cases in 2010–2011, whereas the proportion admitted to ICU decreased from 63% in 2001–2002 to 39% in 2010–2011 (Table 1). TABLE 1. Pattern of Hospitalized Road Transport–Related Major Trauma (ISS >12) in Victoria by Year

Risk-Adjusted In-hospital Mortality The pattern of hospitalized road transport–related major trauma changed over time (Table 1). The mean age of injured patients increased, as did the proportion of cases with substantial comorbidities. There was no clear change in the ISS of patients, but the proportion with an intracranial injury (AIS head Injury Severity Score ≤2) decreased over time (Table 1). Direct transport from the scene of injury to a major trauma service increased from 55% of cases in 2001–2002 to 77% of cases in 2010–2011, whereas the proportion admitted to ICU decreased from 63% in 2001–2002 to 39% in 2010–2011 (Table 1). TABLE 1. Pattern of Hospitalized Road Transport–Related Major Trauma (ISS >12) in Victoria by Year 2001—2002 2002—2003 2003–2004 2004–2005 2005–2006 2006–2007 2007–2008 2008–2009 2009–2010 2010–2011 Descriptor (n = 663) (n = 578) (n = 714) (n = 711) (n = 824) (n = 844) (n = 804) (n = 896) (n = 857) (n = 937) Age, mean (SD) 35.7 (19.0) 37.1 (21.0) 38.7 (20.0) 38.8 (20.3) 39.2 (20.4) 37.5 (19.5) 39.4 (20.2) 40.3 (20.6) 42.3 (21.2) 41.9 (20.8) Sex, n (%) male 467 (70.4) 408 (70.6) 489 (68.5) 479 (67.4) 567 (68.8) 597 (70.7) 630 (69.7) 635 (70.9) 585 (68.3) 706 (75.4) CCI weight, n (%) 0 553 (83.4) 440 (76.1) 534 (74.8) 523 (73.6) 596 (72.3) 590 (69.9) 497 (61.8) 564 (62.9) 532 (62.1) 626 (66.8) 1 86 (13.0) 95 (16.5) 139 (19.5) 141 (19.8) 171 (20.8) 212 (25.1) 262 (32.6) 259 (28.9) 265 (30.9) 230 (24.6) >1 24 (3.6) 43 (7.4) 41 (5.7) 47 (6.6) 57 (6.9) 42 (5.0) 45 (5.6) 73 (8.2) 60 (7.0) 81 (8.6) Road user group, n (%) Motor vehicle 406 (61.2) 365 (63.1) 424 (59.4) 432 (60.7) 493 (59.8) 487 (57.7) 473 (58.8) 486 (54.2) 464 (54.1) 532 (56.8) Motorcycle 119 (18.0) 81 (14.0) 119 (16.7) 118 (16.6) 139 (16.9) 159 (18.8) 158 (19.7) 190 (21.2) 171 (20.0) 172 (18.3) Pedestrian 105 (15.8) 98 (17.0) 115 (16.1) 120 (16.9) 135 (16.4) 127 (15.1) 111 (13.8) 131 (14.6) 125 (14.6) 132 (14.1) Other road user 33 (5.0) 34 (5.9) 56 (7.8) 41 (5.8) 57 (6.9) 71 (8.4) 62 (7.7) 89 (9.9) 97 (11.3) 101 (10.8) Direct transport to major trauma service, n (%) Yes 363 (54.8) 329 (56.9) 468 (65.6) 494 (69.5) 579 (70.3) 626 (74.2) 573 (71.3) 631 (70.4) 618 (72.1) 717 (76.5) ISS group 13–15 139 (21.0) 153 (26.5) 177 (24.8) 169 (23.8) 189 (22.9) 178 (21.1) 193 (24.0) 197 (22.0) 233 (27.2) 250 (26.7) 16–19 163 (24.6) 146 (25.2) 187 (26.2) 184 (25.9) 217 (26.3) 218 (25.8) 204 (25.4) 247 (27.6).

329 (56.9) 468 (65.6) 494 (69.5) 579 (70.3) 626 (74.2) 573 (71.3) 631 (70.4) 618 (72.1) 717 (76.5) ISS group 13–15 139 (21.0) 153 (26.5) 177 (24.8) 169 (23.8) 189 (22.9) 178 (21.1) 193 (24.0) 197 (22.0) 233 (27.2) 250 (26.7) 16–19 163 (24.6) 146 (25.2) 187 (26.2) 184 (25.9) 217 (26.3) 218 (25.8) 204 (25.4) 247 (27.6). 214 (25.0) 237 (25.3) 20–28 188 (28.3) 142 (24.6) 168 (23.5) 161 (22.6) 195 (23.7) 217 (25.7) 195 (24.2) 232 (25.9) 228 (26.6) 242 (25.8) >28 173 (26.1) 137 (23.7) 182 (25.5) 197 (27.7) 223 (27.1) 231 (27.4) 212 (26.4) 220 (24.5) 182 (21.2) 208 (22.2) Head injury severity score, n (%) AIS <3 350 (52.8) 310 (53.6) 388 (54.3) 412 (58.0) 512 (62.1) 438 (51.9) 458 (57.0) 517 (57.7) 509 (59.4) 599 (63.9) AIS 3–4 244 (36.8) 208 (36.0) 262 (36.7) 222 (31.2) 243 (29.5) 323 (38.3) 255 (31.7) 299 (33.4) 265 (30.9) 260 (27.8) AIS 5–6 69 (10.4) 60 (10.4) 64 (9.0) 77 (10.8) 69 (8.4) 83 (9.8) 91 (11.3) 80 (8.9) 83 (9.7) 78 (8.3) ICU* stay, n (%) Yes 427 (64.9) 358 (62.3) 420 (59.1) 435 (61.4) 451 (54.9) 480 (57.0) 428 (53.2) 439 (49.1) 416 (48.7) 459 (49.1) Hospital length of stay† Median (IQR) 12 (6–20) 11 (7–22) 11 (6–20) 10 (6–20) 10 (6–17) 10 (5–18) 9 (5–15) 9 (5–17) 9 (5–16) 9 (5–16) In-hospital death, n (%) Yes 95 (14.3) 76 (13.2) 61 (8.5) 74 (10.4) 81 (9.8) 81 (9.6) 74 (9.2) 72 (8.0) 67 (7.8) 77 (8.2) *Data missing for n = 20 cases. †Data missing for n = 305 cases. TABLE 2. Disability Weights and Duration of Disability by Injury Group (n = 3170)

214 (25.0) 237 (25.3) 20–28 188 (28.3) 142 (24.6) 168 (23.5) 161 (22.6) 195 (23.7) 217 (25.7) 195 (24.2) 232 (25.9) 228 (26.6) 242 (25.8) >28 173 (26.1) 137 (23.7) 182 (25.5) 197 (27.7) 223 (27.1) 231 (27.4) 212 (26.4) 220 (24.5) 182 (21.2) 208 (22.2) Head injury severity score, n (%) AIS <3 350 (52.8) 310 (53.6) 388 (54.3) 412 (58.0) 512 (62.1) 438 (51.9) 458 (57.0) 517 (57.7) 509 (59.4) 599 (63.9) AIS 3–4 244 (36.8) 208 (36.0) 262 (36.7) 222 (31.2) 243 (29.5) 323 (38.3) 255 (31.7) 299 (33.4) 265 (30.9) 260 (27.8) AIS 5–6 69 (10.4) 60 (10.4) 64 (9.0) 77 (10.8) 69 (8.4) 83 (9.8) 91 (11.3) 80 (8.9) 83 (9.7) 78 (8.3) ICU* stay, n (%) Yes 427 (64.9) 358 (62.3) 420 (59.1) 435 (61.4) 451 (54.9) 480 (57.0) 428 (53.2) 439 (49.1) 416 (48.7) 459 (49.1) Hospital length of stay† Median (IQR) 12 (6–20) 11 (7–22) 11 (6–20) 10 (6–20) 10 (6–17) 10 (5–18) 9 (5–15) 9 (5–17) 9 (5–16) 9 (5–16) In-hospital death, n (%) Yes 95 (14.3) 76 (13.2) 61 (8.5) 74 (10.4) 81 (9.8) 81 (9.6) 74 (9.2) 72 (8.0) 67 (7.8) 77 (8.2) *Data missing for n = 20 cases. †Data missing for n = 305 cases. TABLE 2. Disability Weights and Duration of Disability by Injury Group (n = 3170) Disability Weights (SE) Injury Group n 6 mo 12 mo 24 mo Final Short-Term Weight Final Long-Term Weight % (95% CI) With Disability at 24 mo Isolated head injury 623 0.245 (0.017) 0.238 (0.016) 0.272 (0.016) 0.257 0.272 63.8 (59.3–68.3) Head and other region injuries 915 0.269 (0.012) 0.247 (0.012) 0.252 (0.012) 0.255 0.252 66.9 (63.5–70.2) Spinal cord injury 108 0.644 (0.042) 0.547 (0.040) 0.550 (0.038) 0.572 0.550 93.6 (88.7–98.6) Spinal column and/or extremity injuries only 140 0.296 (0.031) 0.250 (0.032) 0.215 (0.034) 0.244 0.215 60.7 (50.3–71.2) Chest/abdominal injuries only 91 0.142 (0.033) 0.101 (0.036) 0.107 (0.035) 0.114 0.107 45.3 (34.1–56.6) Chest and spinal column/extremity injuries 630 0.225 (0.014) 0.190 (0.014) 0.174 (0.014) 0.191 0.174 61.4 (57.2–65.6) Other/multitrauma not involving neurotrauma 663 0.257 (0.014) 0.231 (0.013) 0.214 (0.013) 0.229 0.214 64.8 (60.8–68.8) SE indicates standard error.

101 (0.036) 0.107 (0.035) 0.114 0.107 45.3 (34.1–56.6) Chest and spinal column/extremity injuries 630 0.225 (0.014) 0.190 (0.014) 0.174 (0.014) 0.191 0.174 61.4 (57.2–65.6) Other/multitrauma not involving neurotrauma 663 0.257 (0.014) 0.231 (0.013) 0.214 (0.013) 0.229 0.214 64.8 (60.8–68.8) SE indicates standard error. The in-hospital mortality rate for major trauma cases in Victoria was 9.7%. Adjusting for key differences in case-mix over time and known predictors of mortality, there was a significant and sustained reduction in the adjusted odds of mortality after the 2002–2003 financial year (Fig. 2). FIGURE 2. Adjusted odds of mortality for hospitalized road transport–related major trauma in Victoria (July 2001 to June 2011). Disability-Adjusted Life Years Of the 3505 Victorian State Trauma Registry cases eligible for follow-up from January 2008 to December 2010, EQ-5D responses were available from 3170 cases at any of the time points; 2864 at 6 months, 2935 at 12 months and 2771 at 24 months postinjury. The short- and long-term disability weights calculated from the responders to the EQ-5D at follow-up and the proportion of cases still reporting disability at 24 months postinjury are shown in Table 2. The highest disability weight, and prevalence of disability at 24 months, was for spinal cord injury (Table 2).

njury. The short- and long-term disability weights calculated from the responders to the EQ-5D at follow-up and the proportion of cases still reporting disability at 24 months postinjury are shown in Table 2. The highest disability weight, and prevalence of disability at 24 months, was for spinal cord injury (Table 2). Over the decade, the total DALYs attributed to road transport–related serious injury in Victoria was 102,208, which resulted in a total cost of A$5.1 billion using the per capita GDP estimate of a DALY value or A$15.4 billion using the VSLY. There was a 43% reduction in YLLs and a 32% increase in YLDs. The overall result was a 28% reduction in DALYs over the decade (Table 3). The overall cost saving per case in 2010–2011 was A$209,750 (GDP per capita), or A$633,446 (VSLY), when compared with the 2001–2002 financial year (Fig. 3). FIGURE 3. Health loss cost per road transport–related death or major trauma case (A$). TABLE 3. Disability-Adjusted Life Years and Costs of Road Transport–Related Deaths and Major Trauma (July 2001 to June 2011)

Over the decade, the total DALYs attributed to road transport–related serious injury in Victoria was 102,208, which resulted in a total cost of A$5.1 billion using the per capita GDP estimate of a DALY value or A$15.4 billion using the VSLY. There was a 43% reduction in YLLs and a 32% increase in YLDs. The overall result was a 28% reduction in DALYs over the decade (Table 3). The overall cost saving per case in 2010–2011 was A$209,750 (GDP per capita), or A$633,446 (VSLY), when compared with the 2001–2002 financial year (Fig. 3). FIGURE 3. Health loss cost per road transport–related death or major trauma case (A$). TABLE 3. Disability-Adjusted Life Years and Costs of Road Transport–Related Deaths and Major Trauma (July 2001 to June 2011) Summary Measure Deaths and In-Hospital Major Trauma Survivors (N) YLLs YLDs DALYs % YLD per Case YLD per Survivor DALY per Case Total Cost—GDP per Capita (million A$) Cost per Case—GDP per Capita (A$) Total Cost—VSLY (million A$) Cost per Case—VSLY (million A$) Financial Year 2001–2002 1022 9788 2456 12,244 20.1 4.3 12.0 612.2 599,018 1849.8 1.81 2002–2003 873 8363 2116 10,479 20.2 4.2 12.0 524.0 600,206 1582.4 1.81 2003–2004 988 7182 2639 9821 26.9 4.0 9.9 491.1 497,027 1483.0 1.50 2004–2005 982 7390 2651 10,041 26.4 4.2 10.2 502.0 511,234 1516.1 1.54 2005–2006 1084 7626 2912 10,538 27.6 3.9 9.7 526.9 486,050 1591.2 1.47 2006–2007 1108 7626 3124 10,750 29.1 4.1 9.7 537.5 485,127 1623.3 1.47 2007–2008 1072 7352 2888 10,240 28.2 4.0 9.6 512.0 477,629 1546.3 1.44 2008–2009 1147 6962 3227 10,189 31.7 3.9 8.9 509.4 444,133 1538.5 1.34 2009–2010 1094 6138 2923 9061 32.3 3.7 8.3 453.1 414,133 1368.2 1.25 2010–2011 1136 5612 3233 8845 36.5 3.8 7.8 442.2 389,268 1335.5 1.18 Overall 10,506 74,039 28,169 102,208 27.6 4.0 9.7 5,110.4 486,425 15,433.3 1.47 DISCUSSION Our study investigated the burden of road transport–related trauma over a 10-year time frame using a variety of measures of mortality and morbidity. During this period, there was a significant reduction in the incidence of death and an increase in the incidence of hospitalized major trauma. There was a rapid and sustained reduction in risk-adjusted mortality for hospitalized road-related major trauma, and the overall DALY burden of serious injury fell by 28%. The estimated annual cost of health loss, based on the VSLY, decreased from $1.85 billion to $1.34 billion. The finding of a reduction in risk-adjusted mortality for hospitalized patients, and overall reductions in deaths, is indicative of the positive contribution of the trauma system to reducing road transport–related injury burden in Victoria.

health loss, based on the VSLY, decreased from $1.85 billion to $1.34 billion. The finding of a reduction in risk-adjusted mortality for hospitalized patients, and overall reductions in deaths, is indicative of the positive contribution of the trauma system to reducing road transport–related injury burden in Victoria. Nathens et al7 studied the association between implementation of an organized trauma system on mortality after motor vehicle crashes across all states in the United States from 1979 to 1995. These authors found a reduction in overall mortality rates after system introduction, but the effect did not appear for 10 years, suggesting that the results related to maturation of the trauma system and refinement of policies and referral patterns.7 We observed a reduction in risk-adjusted mortality shortly after complete implementation of the system and evidence of sustained reduction since the 2003–04 financial year. A feature of the Victorian State Trauma System, which possibly shortened the time frame for impact in effect, was the implementation of center designation concurrently with trauma triage and transfer guidelines. During the study, there was a 20% increase in patients transferred directly to a trauma center from the scene of injury. This has previously been associated with reduced mortality risk.35

me frame for impact in effect, was the implementation of center designation concurrently with trauma triage and transfer guidelines. During the study, there was a 20% increase in patients transferred directly to a trauma center from the scene of injury. This has previously been associated with reduced mortality risk.35 Increased survival could, potentially, increase the nonfatal burden, as more survivors live with impaired health. However, we found that although the overall number of road transport–related major trauma patients rose over the decade and the proportion of DALYs attributed to YLDs increased, the burden per case was lower. This finding reflects the reduction in YLLs, a major contributor to DALY calculations, and a shift toward an ageing cohort of patients where life expectancy with disability is shorter. Previous studies have shown improved functional outcomes for patients managed at trauma centers,12,13 further supporting the potential for trauma systems to reduce overall burden of injury.

ontributor to DALY calculations, and a shift toward an ageing cohort of patients where life expectancy with disability is shorter. Previous studies have shown improved functional outcomes for patients managed at trauma centers,12,13 further supporting the potential for trauma systems to reduce overall burden of injury. The impact of the reduction in DALYs was a substantial reduction in the cost of health loss related to road transport–related trauma in Victoria, ranging from A$209,750 per case using the GDP per capita approach to A$633,446 per case using the VSLY approach. The lower GDP per capita amount is often criticized for equating the value of life with the value of production, and in the transport literature, there is a clear preference for the use of the VSLY.36–38 The VSLY approach estimates the willingness to pay to reduce the risk of death by a known probability and extrapolates to infer the willingness to pay for a “statistical life.” In principle, the approach permits the inclusion of all relevant elements of well-being in the estimated VSLY, potentially providing a better estimate of the overall reduction in burden.

to pay to reduce the risk of death by a known probability and extrapolates to infer the willingness to pay for a “statistical life.” In principle, the approach permits the inclusion of all relevant elements of well-being in the estimated VSLY, potentially providing a better estimate of the overall reduction in burden. Key strengths of this study were the inclusion of multiple burden measures, and the use of population-level datasets, to provide a comprehensive overview of the impact of regionalization of trauma care on injury burden in a defined trauma setting. Nevertheless, there were limitations. Trauma systems are multifaceted, involving primary, secondary, and tertiary prevention.6 Given the observational nature of the study, attributing the gains seen to particular aspects of the trauma system was not possible. Furthermore, calculation of the disability weights involved a cohort of patients at the latter end of the study time frame. Although follow-up rates were high, the potential for biased estimates of the disability weights due to loss to follow-up remains, although a previous study showed little to no impact on annualized disability weights using different approaches to handling loss to follow-up.26 Applying the weights to all years of the study ensured that the effect of any bias was consistent across the study time frame, but removed the potential to establish whether the disability weights changed over time. Finally, although estimates of the cost of health loss were determined using a dollar cost per DALY approach, comprehensive measures of the direct and indirect costs were not calculated.

was consistent across the study time frame, but removed the potential to establish whether the disability weights changed over time. Finally, although estimates of the cost of health loss were determined using a dollar cost per DALY approach, comprehensive measures of the direct and indirect costs were not calculated. CONCLUSIONS Since the introduction of a regionalized, inclusive trauma system in Victoria, Australia, there has been a significant decline in the incidence of mortality, reduced risk-adjusted mortality for hospitalized road transport–related major trauma patients and an overall reduction in burden related to road transport–related injury as measured by DALYs. Although the number of hospitalized transport-related major trauma cases has increased over time, disability burden per case has declined, suggesting that an increase in lives saved does not necessarily result in an overall increase in nonfatal injury burden. The results of this study contribute to a growing evidence base that implementation of inclusive trauma systems can play an important role in reducing the population burden of road traffic injury.

d, suggesting that an increase in lives saved does not necessarily result in an overall increase in nonfatal injury burden. The results of this study contribute to a growing evidence base that implementation of inclusive trauma systems can play an important role in reducing the population burden of road traffic injury. ACKNOWLEDGMENTS The authors thank the National Coroners' Information Service, Sue McLellan, and Steven Macey for their assistance with the data and DALY calculations. The authors also thank David Attwood from the Transport Accident Commission for his suggestions and advice, the data collectors and participating hospitals of the VSTR, and the members of the VSTR Steering Committee: Sylvia Supelak, Sue O'Sullivan, Karen Smith, David Attwood, Diane Gill, Michael Geluk, Ben McKenzie, Bruce Bartley, Peter Trethewey, Marcus Kennedy, Jennie Ponsford, Tony Walker, and Simon Young.

ons and advice, the data collectors and participating hospitals of the VSTR, and the members of the VSTR Steering Committee: Sylvia Supelak, Sue O'Sullivan, Karen Smith, David Attwood, Diane Gill, Michael Geluk, Ben McKenzie, Bruce Bartley, Peter Trethewey, Marcus Kennedy, Jennie Ponsford, Tony Walker, and Simon Young. Disclosure: The Victorian State Trauma Registry (VSTR) is a Department of Health, State Government of Victoria and Transport Accident Commission. Peter Cameron and Belinda Gabbe were supported by a Practitioner Fellowship, and a Career Development Fellowship, from the National Health and Medical Research Council (NHMRC) of Australia, respectively. Ronan Lyons leads 1 of the 4 UK e-health Informatics Research Centres funded by a joint investment from: Arthritis Research UK, the British Heart Foundation, Cancer Research UK, the Chief Scientist Office (Scottish Government Health Directorates), the Economic and Social Research Council, the Engineering and Physical Sciences Research Council, the Medical Research Council, the National Institute for Health Research, the National Institute for Social Care and Health Research (Welsh Government) and the Wellcome Trust (grant reference: MR/K006525/1). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors declare no conflicts of interest.

INTRODUCTION Hepatic resection (HR) still stays as one of the main treatment modalities for the hepatocellular carcinoma (HCC) with or without liver cirrhosis. However, HR should be carefully selected among the patients with well-preserved liver function to avoid the postoperative complications related to liver failure. Also, well-preserved or compensated liver function has been defined by the absence of clinically relevant portal hypertension (PHT). Liver transplantation or nonsurgical treatments are recommended for the HCC patients with the evidence of PHT, rather than HR. The clinical evidence of PHT could be evaluated by various methods, such as hepatic venous pressure gradient (HVPG), dye-retention test, Child-Pugh scoring system, serum routine chemistry, and platelet counts.

ation or nonsurgical treatments are recommended for the HCC patients with the evidence of PHT, rather than HR. The clinical evidence of PHT could be evaluated by various methods, such as hepatic venous pressure gradient (HVPG), dye-retention test, Child-Pugh scoring system, serum routine chemistry, and platelet counts. Measuring HVPG has been known as the most reliable clinical method to estimate portal pressure, and HVPG exceeding 10 mm Hg is defined as decompensated PHT and associated with increased cirrhosis-related complications.1,2 Previous studies reported that HR for the patients with HVPG over 10 mm Hg is associated with serious postoperative complications of liver failure and mortality.3,4 Barcelona-Clinic Liver Cancer group treatment guideline for HCC recommended liver transplantation or nonsurgical modality for the patients with high HVPG ≥ 10 mm Hg.5,6 However, the measurement of HVPG has not been preferred in Asian countries because it requires more complicated procedures than other methods. Required procedures such as puncture of central vein and catheter insertion into vena cava may cause possible complications of bleeding, soft tissue hematoma, nerve injury, or arrhythmia.7,8

measurement of HVPG has not been preferred in Asian countries because it requires more complicated procedures than other methods. Required procedures such as puncture of central vein and catheter insertion into vena cava may cause possible complications of bleeding, soft tissue hematoma, nerve injury, or arrhythmia.7,8 In the East Asian countries like Korea and Japan, the indocyanine green retention test (ICG-R15) has been commonly used for the surgical patients to determine PHT, and extent of hepatectomy.5,9 Measuring ICG-R15 has an advantage over HVPG or direct measurement of portal pressure because it is an easy and safe serological test without procedure-related complications. Previous studies reported that ICG-R15 was a valuable assessment tool to decide on the resectability and extent of hepatectomy.9–11 The value of ICG-R15 20% was supposed to be 10 mm Hg of HVPG, and ICG-R15 exceeding 20% has been considered as clinically relevant PHT.3,12,13 However, there was no study which tried to determine an accurate correlation between ICG-R15 and HVPG. Moreover, the value of ICG-R15 could be inaccurate under various patients’ conditions with jaundice, administration of H2 blocker, and genetic defect of ICG excretion.14–16

onsidered as clinically relevant PHT.3,12,13 However, there was no study which tried to determine an accurate correlation between ICG-R15 and HVPG. Moreover, the value of ICG-R15 could be inaccurate under various patients’ conditions with jaundice, administration of H2 blocker, and genetic defect of ICG excretion.14–16 Abnormal liver function test (LFT) and complete blood cell count (CBC) can be seen in patients with liver dysfunction or cirrhosis. The value of serum bilirubin, serum albumin, prothrombin time (PT-INR), and platelet count are regarded to have correlation with liver cirrhosis.13,17–19 Those are simple and least invasive but have a limited value to stratify the degree of portal pressure.6,9,13 Furthermore, the values of those tests could be easily affected by conditions, such as hydration and nutritional status. Authors of this study attempted to establish a reliable equation of calculated HVPG (cHVPG). First, we analyzed the serological tests, which had the significant correlation with measured HVPG and established a model of cHVPG using the values of those serological tests by the linear regression analysis. Then we prospectively assessed the feasibility of cHVPG for the surgical patients with borderline liver reserve.

st, we analyzed the serological tests, which had the significant correlation with measured HVPG and established a model of cHVPG using the values of those serological tests by the linear regression analysis. Then we prospectively assessed the feasibility of cHVPG for the surgical patients with borderline liver reserve. PATIENTS AND METHODS Correlation Between Hepatic Venous Pressure Gradient and Serological Tests Between January 2006 and December 2008, 171 consecutive patients who underwent the liver surgery in our institute were included to establish the correlation between HVPG and the serological tests (correlation cohort). The patients with obstructive jaundice were excluded in this study. The 171 patients in the correlation cohort underwent preoperative measurements of HVPG, ICG-R15, CBC, and LFT simultaneously within 2 days before the scheduled liver surgery. A regression analysis was performed between the values of 171 patients’ HVPG and the serological tests to establish the equation of cHVPG. All procedures and tests were performed with the informed consent provided to the patients and their legal guardians. The study was conducted after approval of the institutional review board.

on analysis was performed between the values of 171 patients’ HVPG and the serological tests to establish the equation of cHVPG. All procedures and tests were performed with the informed consent provided to the patients and their legal guardians. The study was conducted after approval of the institutional review board. The reasons of liver surgery for 171 patients in the correlation cohort were described in Table 1. Among 171 patients, 73 (42.7%) had hepatitis B virus-related liver disease, 6 (3.5%) had alcoholic liver disease, 4 had hepatitis C virus-related liver disease, and 4 had cryptogenic liver cirrhosis. Eighty of 171 (46.8%) patients had biopsy-proven liver cirrhosis. The 171 patients underwent HR in 131, live donor liver transplantation (LDLT) in 38, and abdominal exploration in 2 due to peritoneal carcinomatosis. Among 131 patients who underwent HR, 101 (77.1%) underwent major HR (≥2 segment resection), and 30 (22.9%) received monosegmentectomy or wedge resection. Liver specimens of 169 patients who received liver surgery were analyzed histologically by a single pathologist (YB Kim). All 171 patients in the correlation cohort recovered from surgery and were discharged uneventfully.

ent major HR (≥2 segment resection), and 30 (22.9%) received monosegmentectomy or wedge resection. Liver specimens of 169 patients who received liver surgery were analyzed histologically by a single pathologist (YB Kim). All 171 patients in the correlation cohort recovered from surgery and were discharged uneventfully. Preoperative Measurements of Hepatic Venous Pressure Gradient After overnight fasting, the patients had been referred to the interventional radiologists (JW Kim and JH Won) who are exclusively responsible for the hepatic hemodynamic intervention. Under the local anesthesia, a 6-french venous introducer was inserted to the right internal jugular vein by the ultrasonography-guided Seldinger technique. A 5-french ballooning catheter with pressure sensor (C2 Cobra catheter; Torcon NB® Advantage catheter, Cook Medical Inc.) was advanced via the introducer into the right hepatic vein under fluoroscopic control. Free hepatic venous pressure (FHVP) was measured after the pressure was stabilized for 1 minute. Then the right hepatic vein was completely occluded by the catheter balloon to measure wedged hepatic venous pressure (WHVP). When the right hepatic vein was compressed or invaded by hepatic malignancy, the middle hepatic vein or the left hepatic vein was chosen for FHVP and WHVP. After 3 sets of measurements, alternating between FHVP and WHVP, the median value was recorded. HVPG was drawn by subtracting FHVP from WHVP.

enous pressure (WHVP). When the right hepatic vein was compressed or invaded by hepatic malignancy, the middle hepatic vein or the left hepatic vein was chosen for FHVP and WHVP. After 3 sets of measurements, alternating between FHVP and WHVP, the median value was recorded. HVPG was drawn by subtracting FHVP from WHVP. Preoperative Measurement of Indocyanine Green-R15 and Laboratory Tests A bolus of ICG was injected to the patients kept under overnight fasting with a dose of 0.5 mg/kg via the cephalic vein of 1 forearm. Fifteen minutes after the ICG injection, 8 mL of blood was sampled from the other forearm in a heparinized bottle. The injection of ICG and the blood sampling was exclusively conducted by a single technician (DH Kang). The concentration of ICG in the plasma was determined by the spectrophotometry at 805 nm (Libra S12 spectrophotometer, Bichrom Ltd.). The value of ICG-R15 was expressed as the percentage retention at 15 minutes. Laboratory tests included CBC, serum electrolytes, serum bilirubin, serum albumin, aspartate transaminase, alanine transaminase, gamma-glutamyl transpeptidase, serum creatinine, and PT-INR. All blood samples for the laboratory tests were drawn after overnight fasting on the same day of ICG-R15 test.

Preoperative Measurement of Indocyanine Green-R15 and Laboratory Tests A bolus of ICG was injected to the patients kept under overnight fasting with a dose of 0.5 mg/kg via the cephalic vein of 1 forearm. Fifteen minutes after the ICG injection, 8 mL of blood was sampled from the other forearm in a heparinized bottle. The injection of ICG and the blood sampling was exclusively conducted by a single technician (DH Kang). The concentration of ICG in the plasma was determined by the spectrophotometry at 805 nm (Libra S12 spectrophotometer, Bichrom Ltd.). The value of ICG-R15 was expressed as the percentage retention at 15 minutes. Laboratory tests included CBC, serum electrolytes, serum bilirubin, serum albumin, aspartate transaminase, alanine transaminase, gamma-glutamyl transpeptidase, serum creatinine, and PT-INR. All blood samples for the laboratory tests were drawn after overnight fasting on the same day of ICG-R15 test. Clinical Application of Calculated Hepatic Venous Pressure Gradient Between January 2009 and December 2013, we applied the cHVPG by the K-equation for determination of PHT in the surgical patients with HCC (application cohort). In this period, the measurement of HVPG was not performed for preoperative evaluation of HCC, and the resectability of HCC was determined by the value of cHVPG by the K-equation, regardless of single value of serological test. The patient with cHVPG < 10 mm Hg was considered as having no PHT and regarded as a candidate of HR but cHVPG ≥ 10 mm Hg was considered as an evidence of PHT and unsuitable for HR. The patients with evident PHT with cHVPG ≥ 10 mm Hg were recommended to perform treatments other than HR such as liver transplantation, local ablation, or transcatheter arterial chemoembolization. To evaluate reliability of the cHVPG for HR, we divided the patients with cHVPG < 10 mm Hg (candidate of HR) into 2 groups by the value of ICG-R15 20%. Among the surgical patients with cHVPG < 10 mm Hg, the patients with ICG-R15 < 20% were regarded as group A and ICG-R15 ≥ 20% as group B. Operative outcomes and postoperative complications of group A and B patients were recorded. The short-term and long-term outcomes of HR in group A and B patients were analyzed and compared to determine the clinical feasibility of cHVPG in the assessment of PHT for surgical patients.

roup A and ICG-R15 ≥ 20% as group B. Operative outcomes and postoperative complications of group A and B patients were recorded. The short-term and long-term outcomes of HR in group A and B patients were analyzed and compared to determine the clinical feasibility of cHVPG in the assessment of PHT for surgical patients. Statistical Analysis Statistical analysis was performed using SPSS statistics 13.0. Data were expressed as mean or median values, ranges, and percentages. Univariate analysis was performed by the Student t test or χ2 test. In multivariate analysis, the value of HVPG was correlated to the serological tests using the linear regression analysis. The survival rates were analyzed by the Kaplan-Meier test. P values <0.05 were regarded as the valid significance in statistics. The value of R2, by the regression analysis, determined the reliability of the regression equation. The fitness of the regression model has been validated by the residual plots and analysis of variance (ANOVA) with F-statistics. Sensitivity and specificity for the potential diagnostic performance to predict PHT by cHVPG were assessed by receiver operating characteristic (ROC) curve.

bility of the regression equation. The fitness of the regression model has been validated by the residual plots and analysis of variance (ANOVA) with F-statistics. Sensitivity and specificity for the potential diagnostic performance to predict PHT by cHVPG were assessed by receiver operating characteristic (ROC) curve. RESULTS In 171 patients of the correlation cohort, preoperative HVPG and the serological tests were measured without serious complications. However, 3 of 171 (1.8%) developed soft tissue hematoma related with central venous puncture of the right jugular vein. The cervical hematoma was managed conservatively and resolved without sequelae. The patients’ characteristics, measurement of HVPG, the serological tests, and liver histology of the correlation cohort were described in Table 2.

veloped soft tissue hematoma related with central venous puncture of the right jugular vein. The cervical hematoma was managed conservatively and resolved without sequelae. The patients’ characteristics, measurement of HVPG, the serological tests, and liver histology of the correlation cohort were described in Table 2. Correlation of Hepatic Venous Pressure Gradient With Serological Test The 171 patients in the correlation cohort showed mean 6.72 ± 5.9 mm Hg of HVPG (range from 0 to 31 mm Hg, 95% confidence interval of 0.8–12.6 mm Hg). Among them, 129 (75%) showed HVPG less than 10 mm Hg, and remaining 42 had HVPG ≥ 10 mm Hg. The 129 patients with HVPG < 10 mm Hg in the correlation cohort received right trisectionectomy in 2, right hemihepatectomy in 30, right anterior or posterior sectionectomy in 32, left hemihepatectomy in 29, left lateral sectionectomy in 8, monosegmentectomy in 10, wedge resection in 12, exploratory laparotomy only in 2, and LDLT in 4. The 42 patients with HVPG ≥ 10 mm Hg underwent LDLT (n = 34), and wedge resection (n = 8). The value of HVPG was significantly correlated with ICG-R15, platelet count, PT-INR, serum total bilirubin, and serum albumin in univariate analysis. Multivariate analysis revealed that the value of ICG-R15, platelet count, PT-INR, and serum albumin were independently correlated to HVPG; however, total bilirubin had no significant correlation (P = 0.652). The histological examination of the surgical specimens showed that only stage IV liver fibrosis was significantly correlated with HVPG ≥ 10 mm Hg (Table 3).

e of ICG-R15, platelet count, PT-INR, and serum albumin were independently correlated to HVPG; however, total bilirubin had no significant correlation (P = 0.652). The histological examination of the surgical specimens showed that only stage IV liver fibrosis was significantly correlated with HVPG ≥ 10 mm Hg (Table 3). In the correlation cohort, ICG-R15 value was mean 19.7 ± 13.8 with ranged from 3.6% to 65.2%, serum albumin was mean 3.84 ± 0.62 (range from 2.3 to 5.2 g/dL), platelet count was mean 207,098 ± 100,235 per μL (range from 27,000 per to 575,000 per μL), and PT-INR was mean 1.24 ± 0.5 (range from 0.84 to 4.59). The individual correlation of HVPG with ICG-R15, serum albumin, platelet count, and PT-INR by a linear regression analysis was shown in Figure 1. The coefficient of determination (R2 value) was expressed for each correlation. FIGURE 1 Correlation of HVPG with serological tests in the correlation cohort. Hepatic venous pressure gradient was significantly correlated with serological tests of ICG-R15, serum albumin, platelet count, and prothrombin time by multivariate analysis. Among them, the ICG-R15 had the most significant correlation with HVPG (R2 = 0.656). ICG-R15 (A) and prothrombin time-international normalized ratio (D) were positively correlated to HVPG (P < 0.05), and serum albumin level (B) and platelet count (C) are inversely correlated to HVPG (P < 0.05). A total of 95% prediction intervals are shown as dashed lines. Locally weighted scatter plot smooth for each serological test is shown as dotted line.

ional normalized ratio (D) were positively correlated to HVPG (P < 0.05), and serum albumin level (B) and platelet count (C) are inversely correlated to HVPG (P < 0.05). A total of 95% prediction intervals are shown as dashed lines. Locally weighted scatter plot smooth for each serological test is shown as dotted line. The equations showed that ICG-R15 and PT-INR were directly correlated to HVPG but platelet count and serum albumin level were inversely correlated. According to the R2 values of the equations, ICG-R15 had the best reliable correlation with HVPG among the serological tests (R2 = 0.656). According to the above equation of ICG-R15, 25.8% of ICG-R15 value corresponded to the value of 10 mm Hg in terms of HVPG. Additionally with the above equations, HVPG 10 mm Hg was equivalent to 3.3 g/dL of serum albumin 119,000 per μL of platelet count, and 1.67 of PT-INR. All 4 model assumptions were assessed by ANOVA test with F-statistics (P < 0.05). By multivariate linear regression analysis, the cHVPG was established as following equation (K-equation).

The equations showed that ICG-R15 and PT-INR were directly correlated to HVPG but platelet count and serum albumin level were inversely correlated. According to the R2 values of the equations, ICG-R15 had the best reliable correlation with HVPG among the serological tests (R2 = 0.656). According to the above equation of ICG-R15, 25.8% of ICG-R15 value corresponded to the value of 10 mm Hg in terms of HVPG. Additionally with the above equations, HVPG 10 mm Hg was equivalent to 3.3 g/dL of serum albumin 119,000 per μL of platelet count, and 1.67 of PT-INR. All 4 model assumptions were assessed by ANOVA test with F-statistics (P < 0.05). By multivariate linear regression analysis, the cHVPG was established as following equation (K-equation). K-equation of multivariate analysis using significant correlation factors had coefficient of determination (0.707 of adjusted R2 value). The validity of K-equation has been assessed by the residual plots and ANOVA test (F = 98.278, P < 0.001). Cross-validation was also performed to estimate model (K-equation) performance. The 10-fold cross-validation performed allowing nonzero intercept term showed 0.661 of R2 value. The correlation plots between HVPG and cHVPG are shown in Figure 2. With the ROC curve for the correlation cohort, cHVPG values were plotted for their ability to predict the patients to be without PHT (HVPG < 10 mm Hg). The area under the curve of the ROC curve was 0.96 (95% confidence interval; 0.93–0.99, P < 0.001). At the point on the plot where a patient with cHVPG < 10 mm Hg, sensitivity was 98.4% and specificity was 76.2%, respectively, for identification of the patients without PHT. Also, the positive predictive value was 92.7% and the negative predictive value was 94.1% to predict the patients to be without PHT (Fig. 3).

1). At the point on the plot where a patient with cHVPG < 10 mm Hg, sensitivity was 98.4% and specificity was 76.2%, respectively, for identification of the patients without PHT. Also, the positive predictive value was 92.7% and the negative predictive value was 94.1% to predict the patients to be without PHT (Fig. 3). FIGURE 2 The scatterplots between hepatic venous pressure gradient and calculated hepatic venous pressure gradient. A total of 95% prediction intervals were shown as dashed lines. FIGURE 3 Receiver operating characteristics curve for calculated hepatic venous pressure gradient values in the correlation cohort to predict absence of portal hypertension (hepatic venous pressure gradient <10 mm Hg). A, The maximal value of Youden index. B, P < 0.001.

FIGURE 2 The scatterplots between hepatic venous pressure gradient and calculated hepatic venous pressure gradient. A total of 95% prediction intervals were shown as dashed lines. FIGURE 3 Receiver operating characteristics curve for calculated hepatic venous pressure gradient values in the correlation cohort to predict absence of portal hypertension (hepatic venous pressure gradient <10 mm Hg). A, The maximal value of Youden index. B, P < 0.001. Clinical Application of Calculated Hepatic Venous Pressure Gradient by K-equation Between January 2009 and December 2013, 510 consecutive HCC patients of the application cohort underwent the surgical evaluation in our institute. Among them, 17 patients were discovered that they had extra-hepatic metastasis of HCC during surgical evaluation, and they were converted to have palliative treatment. Remaining 493 surgical patients with HCC in the application cohort had further evaluation of resectability with the preoperative value of cHVPG. Among them, 452 patients had cHVPG < 10 mm Hg and were considered as candidates of HR, and finally 425 received HR and 27 received LDLT. These 27 LDLT patients were excluded from further analysis in the application cohort. Among the 425 patients who received HR, 357 patients had preoperative ICG-R15 < 20% (group A) and 68 had ICG-R15 ≥ 20% (group B). There was no patient with ICG-R15 < 20% who had cHVPG of more than 10 mm Hg. The remaining 41 patients, who had preoperative cHVPG ≥ 10 mm Hg, received nonresective treatments. The 41 patients with cHVPG ≥ 10 mm Hg showed 13.35 mm Hg of median cHVPG (ranged from 10.17 to 25.29 mm Hg) and 39.8% of median ICG-R15 (ranged from 24.7% to 64.4%). The treatment flow chart of 510 patients in the application cohort was shown in Figure 4.

preoperative cHVPG ≥ 10 mm Hg, received nonresective treatments. The 41 patients with cHVPG ≥ 10 mm Hg showed 13.35 mm Hg of median cHVPG (ranged from 10.17 to 25.29 mm Hg) and 39.8% of median ICG-R15 (ranged from 24.7% to 64.4%). The treatment flow chart of 510 patients in the application cohort was shown in Figure 4. FIGURE 4 Treatment flow chart of the hepatocellular carcinoma patients in the application cohort by the cHVPG. The surgical patients with hepatocellular carcinoma in the application cohort were evaluated and selected for hepatic resection according to cHVPG. The hepatic resection was considered for the patients with cHVPG < 10 mm Hg. The hepatic resection was not recommended to the patient with cHVPG ≥ 10 mm Hg (n = 41). The hepatic resection was performed for the patients in group A and group B regardless of indocyanine green retention test.

resection according to cHVPG. The hepatic resection was considered for the patients with cHVPG < 10 mm Hg. The hepatic resection was not recommended to the patient with cHVPG ≥ 10 mm Hg (n = 41). The hepatic resection was performed for the patients in group A and group B regardless of indocyanine green retention test. The preoperative characteristics and operative outcomes of the 425 HR patients were analyzed and compared between group A (n = 357) and group B (n = 68) in Table 4. The mean value of cHVPG of group A patients was 4.76 ± 1.4 and that of group B was 7.80 ± 1.1. The patients’ characteristics showed no statistical difference between 2 groups. The values of serological tests were similar between 2 groups of patients except ICG-R15 (12.6 ± 4.1% in group A vs 24.6 ± 3.7% in group B, P < 0.001). The pathological characteristics of HCC and intraoperative outcomes of 2 groups were not statistically different (P > 0.05). The major HR (≥2 segment resection) was performed in 194 of group A (54%) and 32 of group B (47%), and the detailed type of HR was described and compared in Table 5. The posthepatectomy laboratory findings showed no significant difference between 2 groups in terms of serum bilirubin, PT-INR, serum creatinine, serum albumin, and platelet count. Also, the incidences of surgical complications after HR were similar between 2 groups (P > 0.05). There were 20 patients who drained more than 500 mL of ascites per day after posthepatectomy day 14. They were treated by dietary sodium restriction and diuretics. There was no patient who developed hepatic encephalopathy after HR in the application cohort. There was 1 case of in-hospital mortality among group B patients on posthepatectomy day 29 due to Acinetobacter pneumonia. She was 69 years old and had underlying chronic obstructive pulmonary disease. The length of hospital stay after HR was similar between 2 groups. The remaining 424 patients recovered from HR and were discharged with favorable liver function. There was no 3-month mortality among the 424 patients who were discharged from hospital uneventfully. However, 6 patients died within 6 months after HR due to recurrent HCC (n = 3), liver failure (n = 2), or variceal bleeding (n = 1). During median follow-up periods of 22 months after HR, 177 (41.6%) patients experienced HCC recurrence. However, there was no difference in recurrence of HCC, and 5-year survival rates were similar between 2 groups (P > 0.05).

s after HR due to recurrent HCC (n = 3), liver failure (n = 2), or variceal bleeding (n = 1). During median follow-up periods of 22 months after HR, 177 (41.6%) patients experienced HCC recurrence. However, there was no difference in recurrence of HCC, and 5-year survival rates were similar between 2 groups (P > 0.05). DISCUSSION HR is one of the major curative modality of treatments for HCC. An extensive resection of liver parenchyme up to 70% of total liver volume could be performed safely in the patients with normal background liver histology.20,21 However, HCC is usually developed from liver with chronic disease and cirrhosis. Thus, HR could be applied to the limited number of HCC patients with compensated or well-preserved liver cirrhosis defined by HVPG less than 10 mm Hg.1,2,22 HVPG was reported as the most reliable methods to assess PHT in many study.1,2,22–25 However, the measurement of HVPG has limitations of invasiveness, high cost, and need for skilled radiologist and high-technology facilities. Various serological tests, such as ICG-R15, LFT, and platelet count are widely used as surrogate methods to assess PHT despite of lower reliability than HVPG. Until now, there was no study which reported the relationship between HVPG and the serological tests for portal pressure assessment.

high-technology facilities. Various serological tests, such as ICG-R15, LFT, and platelet count are widely used as surrogate methods to assess PHT despite of lower reliability than HVPG. Until now, there was no study which reported the relationship between HVPG and the serological tests for portal pressure assessment. The present study is the first report of statistical analysis for the quantitative correlation between HVPG and the serological tests for assessment of PHT. In this study, the quantitative correlation between HVPG and the serological tests were derived by univariate and multivariate linear regression analysis of the correlation cohort of 171 surgical patients. The 171 surgical patients for the correlation analysis between HVPG and the serological tests had a spectrum of liver function and histology from live donor's liver to end-stage liver disease. Authors of this study regarded that the wide distribution of portal pressure values due to various status of liver histology in the correlation cohort was suitable for making a good correlation equation to define clinically relevant PHT. The adjusted R2 value of the K-equation was 0.707, which was considered to be favorable reliability for the model, and the coefficient of the level-level regression between cHVPG and HVPG was close to 1 (0.984) as shown in Figure 2.

was suitable for making a good correlation equation to define clinically relevant PHT. The adjusted R2 value of the K-equation was 0.707, which was considered to be favorable reliability for the model, and the coefficient of the level-level regression between cHVPG and HVPG was close to 1 (0.984) as shown in Figure 2. If the patients in the correlation cohort had normal distribution with mean value of 10 mm Hg of HVPG, the correlation cohort could have been statistically ideal, and the correlation between HVPG and serological tests could have been the most helpful for the patients to determine PHT, because 10 mm Hg of HVPG is well-known cut-off value for PHT. However, 171 consecutive patients of the correlation cohort in this study showed the mean HVPG value of 6.7 ± 5.9 mm Hg, and a relatively small portion (25%, n = 42) of patients had HVPG ≥ 10 mm Hg. The reason of this uneven distribution would be that the cohort was composed of surgical candidates, and the number of the patients was not modified intentionally to fit the mean value of HVPG to 10 mm Hg. However, the constructed ROC curve for the correlation cohort in this study showed that the absence or presence of PHT could be predicted by cut-off value, 10 mm Hg of cHVPG, with 92.7% of positive predictive value and 94.1% of negative predictive value (Fig. 3). It could be understood as the predictability of a patient with cHVPG less than 10 mm Hg to have an actual value of HVPG less than 10 mm Hg was 92.7%, which provided a safe selection criterion for HR in this study. In the ROC curve, 7.495 mm Hg of cHVPG was the maximal value of Youden index but an evaluation metrics of kappa statistics for agreement showed that the 10 mm Hg of cHVPG outperformed 7.495 mm Hg of cHVPG to predict PHT.26,27 Therefore, we considered that uneven distribution of the correlation cohort had little hindrance to reliability of cHVPG in predicting PHT.

e maximal value of Youden index but an evaluation metrics of kappa statistics for agreement showed that the 10 mm Hg of cHVPG outperformed 7.495 mm Hg of cHVPG to predict PHT.26,27 Therefore, we considered that uneven distribution of the correlation cohort had little hindrance to reliability of cHVPG in predicting PHT. In the correlation cohort of this study, the assessments of portal pressure were performed within 2 days before the elective surgery, which may strengthen the reliability of the tests by providing homogeneous status of hydration and nutritional support during hospitalization. This could calibrate patients’ factors to measure reliable values of HVPG and the serological tests for portal pressure. Also, we excluded the patients with obstructive jaundice or acute cholestasis such as hilar cholangiocarcinoma from the correlation cohort because those conditions could influence the value of indocyanine test to be abnormally high.28,29 In this study, the measurement of HVPG and ICG-R15 were performed by exclusively responsible experts in procedures. However, 3 patients (1.8%) developed cervical hematoma as a complication of HVPG measurement, managed conservatively with compression dressing. According to our experience, we also acknowledged the invasiveness of HVPG measurement and that it should be performed by skilled hands.

vely responsible experts in procedures. However, 3 patients (1.8%) developed cervical hematoma as a complication of HVPG measurement, managed conservatively with compression dressing. According to our experience, we also acknowledged the invasiveness of HVPG measurement and that it should be performed by skilled hands. Many previous studies reported that the value of HVPG 10 mm Hg could be a cut-off value of clinical relevant PHT in chronic liver disease, related to decompensated symptoms of liver cirrhosis.4,13 The patients with decompensated liver cirrhosis may also present abnormal values of the serological tests, such as thrombocytopenia, hyperbilirubinemia, and hypoalbuminemia. Multivariate regression analysis of this study showed that the value of HVPG was significantly correlated with ICG-R15, serum albumin, PT-INR, and platelet count. The serum level of total bilirubin had a correlation with HVPG in the univariate analysis (P = 0.003) but had no significance in the multivariate regression (P = 0.65). However, it was difficult to interpret the reason that the serum bilirubin was significant in the univariate analysis only. Bilirubin alone is neither sensitive nor specific for intrinsic liver disease but serves as an indirect measure of the ability of the liver to take up and conjugate bilirubin and to secrete it eventually.30 This study developed the individual correlation equations between the value of HVPG and the 4 significant serological tests: ICG-R15, serum albumin, PT-INR, and platelet count. The reliability of each equation was determined by R2 value and ANOVA test with F-statistics, which would give information about the goodness of fit and statistical significance of our model. Authors of this study have assumed that linear relationship existed between HVPG and each test based on simplicity because those 4 serological tests have been well-known conventional parameters for predicting PHT. Also, there was no previous report, which suggested true model of their relationship. However, in this study, there seems to be nonlinear relationship between HVPG and some of serological tests, revealed by the local smoothers, especially in Figure 1C and 1D. We acknowledge that the weak linearity could be a potential limitation of this study.

us report, which suggested true model of their relationship. However, in this study, there seems to be nonlinear relationship between HVPG and some of serological tests, revealed by the local smoothers, especially in Figure 1C and 1D. We acknowledge that the weak linearity could be a potential limitation of this study. ICG is a protein-binding anionic organic dye that is selectively taken up by hepatocytes and excreted unchanged via the bile. The removal of ICG reflects the capabilities of the liver to uptake and excrete, which can be extrapolated to reflect hepatocyte blood flow and functional hepatocytes mass. ICG elimination is, by far, the most widely used and published functional assessment of liver reserve worldwide and has also been useful in predicting short-term prognosis in liver transplant patients.30 Conventionally, ICG-R15 value of 20% was regarded as the cut-off value for clinically relevant PHT, and the value was suggested by clinical experience or recommendation of many centers. ICG-R15 value of 20% was supposed to be equivalent to HVPG of 10 mm Hg.9,31 However, the HVPG of 10 mm Hg was quantitatively equivalent to ICG-R15 value of 25.8% by the correlation equation of this study. The difference of cut-off value of ICG-R15 for clinical PHT between the quantitative value of this study and conventional value could have resulted in overestimation of clinical PHT to the patients with ICG-R15 values between 20% and 25.8%. Those patients with HCC could be deprived of the chance to have HR due to overestimated risk of postoperative complication related to liver failure, if the treatment modality for HCC were selected by the result of ICG-R15 test. The quantification of ICG-R15 for PHT in this study may help liver surgeons to assess liver function of surgical patients more precisely and decide the treatment modality more accurately. The serum albumin level, PT-INR, and platelet count were also significantly correlated to HVPG by multivariate analysis. Serum albumin and PT-INR represent synthetic function of the liver, and these have been used in Child-Pugh scoring system for stratification of chronic liver disease. Both serological parameters were produced exclusively by the liver. The serum albumin value of 3.3 g/dL and PT-INR value of 1.67 correspond to HVPG of 10 mm Hg by the individual equations in this study. The values of serum albumin and PT-INR for PHT were similar to the value of decompensated chronic liver disease in Child-Pugh scoring system.

ers were produced exclusively by the liver. The serum albumin value of 3.3 g/dL and PT-INR value of 1.67 correspond to HVPG of 10 mm Hg by the individual equations in this study. The values of serum albumin and PT-INR for PHT were similar to the value of decompensated chronic liver disease in Child-Pugh scoring system. Platelet count of 119,000 per μL was correlated to 10 mm Hg of HVPG by the equation, and the value was similar to the results of previous studies.3,32 The median body mass index (BMI) of the correlation cohort of this study was 23.4 kg/m2, which might be lesser than those of the Western population. However, previous studies from the Western countries reported that the value of ICG-R15 or HVPG had not been significantly affected by BMI.4,33,34 Also, the result of this study was consistent with those of the previous reports. Thus, we consider that the equation of our study could be translated to population with a higher BMI.

revious studies from the Western countries reported that the value of ICG-R15 or HVPG had not been significantly affected by BMI.4,33,34 Also, the result of this study was consistent with those of the previous reports. Thus, we consider that the equation of our study could be translated to population with a higher BMI. We assessed the clinical feasibility of K-equation on a following prospective cohort of 510 surgical patients with HCC. Among the patients in the application cohort of this study, no one with favorable ICG-R15 < 20% showed cHVPG ≥ 10 mm Hg according to K-equation. This result could support the reliability of the correlation equation between HVPG and ICG-R15 of the correlation cohort, which suggested that 10 mm Hg of HVPG was equivalent to 25.8% of ICG-R15. Forty-one patients in the Application cohort who showed the unfavorable value of cHVPG ≥10 mm Hg were recommended to undergo non-HR treatment according to the treatment guideline of the previous study.3,6 Our decision to provide nonresective treatments for those patients with cHVPG ≥10 mm Hg was appropriately supported by their high values of ICG-R15 (median 39.8%, ranged from 24.7% to 64.4%). However, one of the limitations of our study is that clinical validation of K-equation was impossible in the strict sense because data for outcomes of HR for patients with high values of cHVPG (≥10 mm Hg) could not be collected.

ately supported by their high values of ICG-R15 (median 39.8%, ranged from 24.7% to 64.4%). However, one of the limitations of our study is that clinical validation of K-equation was impossible in the strict sense because data for outcomes of HR for patients with high values of cHVPG (≥10 mm Hg) could not be collected. In the application cohort of this study, HR was recommended and performed in patients with cHVPG < 10 mm Hg. There were 27 patients who underwent LDLT for HCC even with cHVPG < 10 mm Hg. In our institute, LDLT could be performed for HCC patients without vascular invasion or metastasis on the familial support of a live donor. The remaining patients with cHVPG < 10 mm Hg who underwent HR could be divided into 2 groups according to ICG-R15 value, either less than 20% (group A, n = 357) or more than or equal to 20% (group B, n = 68). Unless we applied K-equation for cHVPG, the patients of group B might have not been recommended for HR because of unfavorable ICG-R15. The Japanese surgical guideline for HCC recommended that the major HR including 2 or more segments should be performed in patients with ICG-R15 less than 20%, and the policy has been accepted widely in many center.9,35 However, in the application cohort in this study, 32 patients (47%) in group B received major HR successfully, and 22 of them underwent more extensive HR than hemihepatectomy on the basis of favorable results of cHVPG (<10 mm Hg). In this study, there was no significant difference in intraoperative outcomes, types of HR, operative complications, and survival rates after HR between group A and B. The successful surgical outcomes of group B patients in the application cohort of this study could imply that the cHVPG by K-equation should be a clinically reliable model to determine whether a surgical patient has preserved liver function or not. This study has drawn that the cHVPG had higher clinical reliability and accuracy than separate use of the serologic tests for assessment of liver reserve function of the surgical patients. Therefore, authors of this study suggested that K-equation for cHVPG was well established and its clinical feasibility was assessed by the correlation and the application cohorts in this study. The clinical application of K-equation for cHVPG could be useful to determine the evidence of PHT for the surgical patients with HCC. We expect that it is one of the strong merits that the cHVPG can be drawn without invasive procedures.

sibility was assessed by the correlation and the application cohorts in this study. The clinical application of K-equation for cHVPG could be useful to determine the evidence of PHT for the surgical patients with HCC. We expect that it is one of the strong merits that the cHVPG can be drawn without invasive procedures. Assessment of the liver reserve function for the liver surgery using ICG-R15 has been widely performed in the most Eastern centers and some European centers, whereas ICG-R15 test might not be available in many centers in the United States and the West. This could be one of the limitations for wide application of the equation for cHVPG because the ICG-R15 is one of the significant variables in the equation. Besides, the value of ICG-R15 might be affected by some conditions such as jaundice and some genetic disorders.36–38 Uptake of ICG by hepatocytes is regulated by ATP-independent organic anion-transporting polypeptide (OATP) located at the basolateral membrane of the liver. Bilirubin competes with ICG for hepatocyte uptake via OATP. Thus, the HCC patients with jaundice due to common hepatic duct invasion might have abnormally high value of ICG-R15. Also, the patients with genetic defect in OATP like Rotor syndrome and constitutional ICG excretory defect should have unreliable value of ICG-R15. These conditions would be another limitation in application of the equation for the cHVPG, and direct measurement of the HVPG through interventional radiology could be selected to determine PHT for these patients.

OATP like Rotor syndrome and constitutional ICG excretory defect should have unreliable value of ICG-R15. These conditions would be another limitation in application of the equation for the cHVPG, and direct measurement of the HVPG through interventional radiology could be selected to determine PHT for these patients. In conclusion, the HVPG could be predicted with serological tests by the K-equation. The cHVPG by K-equation could provide a more feasible clinical reference to determine evidence of PHT than value of each serological test for the surgical patients with HCC. The authors of this study suggest that clinical application of cHVPG could help liver surgeons to assess portal pressure quantitatively and select HR appropriately for the surgical candidates with HCC. Acknowledgments The authors thank Da-Hye Kang for her excellent technical assistance in measurement of ICG-R15. Authorship contribution: BWK and HJW designed the study. BWK, HJW, HYL, JHW, JK, XGH, JS, JBB, and YBK discussed and conducted the study and reviewed the manuscript. TK, BWK, HYL, and YBK collected and analyzed the data and wrote the manuscript. The authors declare no conflicts of interest. TABLE 1 Causes of Liver Surgery in the Correlation Cohort Diagnosis N/171 (%) Hepatocellular carcinoma 57 (33.3) Living liver donor 38 (22.2) End-stage liver cirrhosis 29 (16.9) Metastatic liver cancer 17 (9.9) Hepatolithiasis 15 (8.8) Cholangiocarcinoma 9 (5.3) Other* 6 (3.5) *Other: 2 gallbladder cancer, 1 cystic adenoma, 1 hepatic adenoma, 1 benign inflammatory hepatic mass, and 1 chronic cholecystitis.

lar carcinoma 57 (33.3) Living liver donor 38 (22.2) End-stage liver cirrhosis 29 (16.9) Metastatic liver cancer 17 (9.9) Hepatolithiasis 15 (8.8) Cholangiocarcinoma 9 (5.3) Other* 6 (3.5) *Other: 2 gallbladder cancer, 1 cystic adenoma, 1 hepatic adenoma, 1 benign inflammatory hepatic mass, and 1 chronic cholecystitis. TABLE 2 Patients’ Characteristics and Liver Function Tests of the Correlation Cohort Variables n = 171 Age (years) 49 (16–84) Male sex 119 (69.6%) Weight (kg) 64.4 ± 9.9 Height (cm) 165.8 ± 7.8 BMI (kg/m2) 23.4 (17.1–33.1) HVPG (mm Hg) 6.72 ± 5.9 ICG-R15% 19.69 ± 13.8 PT-INR 1.24 ± 0.5 Platelet count (×103) per μL 207.1 ± 100 Serum creatinine (mg/dL) 0.85 ± 0.2 BUN (mg/dL) 24.61 ± 17.2 Serum total bilirubin (mg/dL) 1.83 ± 4.4 Serum AST (U/L) 43.90 ± 26.7 Serum ALT (U/L) 44.81 ± 36.1 Serum γ-GT (U/L) 81.3 ± 34.2 Serum albumin (g/dL) 3.84 ± 0.6 Liver histology* Steatosis >10% 32 (18.9%) Inflammation 0, I, II/III, IV† 159 (94.1%)/10 (5.9%) Fibrosis 0, I, II, III/IV‡ 89 (52.7%)/80 (47.3%) Data are number (%) or mean ± SD or median (range). ALT, alanine transaminase; AST, aspartate transaminase; BUN, blood urea nitrogen; γ-GT, gamma-glutamyl transpeptidase; ICG-R15%, indocyanine green 15 minutes retention rate; PT-INR, prothrombin time-international normalized ratio. *Liver histology was reviewed in 169 liver specimens. †Inflammation grade: 0, normal; I, minimal; II, mild; III, moderate; and IV, severe. ‡Fibrosis stage: 0, normal; I, portal fibrosis; II, periportal fibrosis; III, septal fibrosis; and IV, cirrhosis.

Variables n = 171 Age (years) 49 (16–84) Male sex 119 (69.6%) Weight (kg) 64.4 ± 9.9 Height (cm) 165.8 ± 7.8 BMI (kg/m2) 23.4 (17.1–33.1) HVPG (mm Hg) 6.72 ± 5.9 ICG-R15% 19.69 ± 13.8 PT-INR 1.24 ± 0.5 Platelet count (×103) per μL 207.1 ± 100 Serum creatinine (mg/dL) 0.85 ± 0.2 BUN (mg/dL) 24.61 ± 17.2 Serum total bilirubin (mg/dL) 1.83 ± 4.4 Serum AST (U/L) 43.90 ± 26.7 Serum ALT (U/L) 44.81 ± 36.1 Serum γ-GT (U/L) 81.3 ± 34.2 Serum albumin (g/dL) 3.84 ± 0.6 Liver histology* Steatosis >10% 32 (18.9%) Inflammation 0, I, II/III, IV† 159 (94.1%)/10 (5.9%) Fibrosis 0, I, II, III/IV‡ 89 (52.7%)/80 (47.3%) Data are number (%) or mean ± SD or median (range). ALT, alanine transaminase; AST, aspartate transaminase; BUN, blood urea nitrogen; γ-GT, gamma-glutamyl transpeptidase; ICG-R15%, indocyanine green 15 minutes retention rate; PT-INR, prothrombin time-international normalized ratio. *Liver histology was reviewed in 169 liver specimens. †Inflammation grade: 0, normal; I, minimal; II, mild; III, moderate; and IV, severe. ‡Fibrosis stage: 0, normal; I, portal fibrosis; II, periportal fibrosis; III, septal fibrosis; and IV, cirrhosis. TABLE 3 Correlation of Hepatic Venous Pressure Gradient With the Patients’ Characteristics and Serological Tests

28.9 (14.5) 1.7 Histological findings of Biopsy specimen Well 872 46.7 645 36.6 20.6 396 42.9 389 42.1 1.5 Poor 640 34.3 786 44.6 −21.2 361 39.1 367 39.7 −1.3 Mixed type 354 19.0 329 18.7 0.8 167 18.1 168 18.2 −0.3 Others 0 0.0 2 0.1 −4.8 0 0.0 0 0.0 NA ASA-PS indicates American Society of Anesthesiologists physical s tatus; BMI, Body mass index; SD, Standard deviation. TABLE 2 Operating and Pathological Findings

*Liver histology was reviewed in 169 liver specimens. †Inflammation grade: 0, normal; I, minimal; II, mild; III, moderate; and IV, severe. ‡Fibrosis stage: 0, normal; I, portal fibrosis; II, periportal fibrosis; III, septal fibrosis; and IV, cirrhosis. TABLE 3 Correlation of Hepatic Venous Pressure Gradient With the Patients’ Characteristics and Serological Tests Variables HVPG < 10 mm Hg (n = 129) HVPG ≥ 10 mm Hg (n = 42) Univariate Analysis, P Multivariate Analysis, P Age (years) 49 (16–84) 49 (36–69) 0.281 — Male sex 88 (68.75%) 31 (72.1%) 0.551 — Weight (kg) 63.62 ± 9.2 66.98 ± 9.5 0.118 — Height (cm) 165.66 ± 7.5 166.14 ± 8.7 0.767 — BMI (kg/m2) 23.1 (17.1–33.1) 24.1 (19.7–32.3) 0.068 — ICG-R15% 13.93 ± 13.9 39.11 ± 15.0 <0.001 <0.001 Platelet count (×1000) 241.18 ± 80.7 91.89 ± 72.4 <0.001 0.003 PT-INR 1.06 ± 0.1 1.83 ± 0.7 <0.001 0.014 Serum creatinine (mg/dL) 0.85 ± 0.1 0.83 ± 0.2 0.612 — BUN (mg/dL) 24.61 ± 17.2 26.10 ± 18.0 0.701 — Serum total bilirubin (mg/dL) 0.82 ± 0.3 5.24 ± 8.4 0.003 0.652 Serum AST (U/L) 43.43 ± 26.1 45.67 ± 28.9. 0.367 — Serum ALT (U/L) 45.52 ± 34.2 42.40 ± 41.8 0.277 — Serum γ-GT (U/L) 81.30 ± 34.2 79.8 ± 39.8 0.432 — Serum albumin (g/dL) 4.04 ± 0.4 3.12 ± 0.5 <0.001 0.004 Liver histology* Steatosis >10% 24 (18.9%) 8 (19%) 0.797 Inflammation 0, I, II/III, IV† 122 (96.1%)/5 (3.9%) 37 (88.1%)/5 (11.9%) 0.070 Fibrosis 0, I, II, III/IV‡ 89 (70.1%)/38 (29.9%) 0/42 (100%) <0.001 Data are number (%) or mean ± SD or median (range). Significance was defined as P < 0.05. *Liver histology was reviewed in 169 liver specimens.

Variables HVPG < 10 mm Hg (n = 129) HVPG ≥ 10 mm Hg (n = 42) Univariate Analysis, P Multivariate Analysis, P Age (years) 49 (16–84) 49 (36–69) 0.281 — Male sex 88 (68.75%) 31 (72.1%) 0.551 — Weight (kg) 63.62 ± 9.2 66.98 ± 9.5 0.118 — Height (cm) 165.66 ± 7.5 166.14 ± 8.7 0.767 — BMI (kg/m2) 23.1 (17.1–33.1) 24.1 (19.7–32.3) 0.068 — ICG-R15% 13.93 ± 13.9 39.11 ± 15.0 <0.001 <0.001 Platelet count (×1000) 241.18 ± 80.7 91.89 ± 72.4 <0.001 0.003 PT-INR 1.06 ± 0.1 1.83 ± 0.7 <0.001 0.014 Serum creatinine (mg/dL) 0.85 ± 0.1 0.83 ± 0.2 0.612 — BUN (mg/dL) 24.61 ± 17.2 26.10 ± 18.0 0.701 — Serum total bilirubin (mg/dL) 0.82 ± 0.3 5.24 ± 8.4 0.003 0.652 Serum AST (U/L) 43.43 ± 26.1 45.67 ± 28.9. 0.367 — Serum ALT (U/L) 45.52 ± 34.2 42.40 ± 41.8 0.277 — Serum γ-GT (U/L) 81.30 ± 34.2 79.8 ± 39.8 0.432 — Serum albumin (g/dL) 4.04 ± 0.4 3.12 ± 0.5 <0.001 0.004 Liver histology* Steatosis >10% 24 (18.9%) 8 (19%) 0.797 Inflammation 0, I, II/III, IV† 122 (96.1%)/5 (3.9%) 37 (88.1%)/5 (11.9%) 0.070 Fibrosis 0, I, II, III/IV‡ 89 (70.1%)/38 (29.9%) 0/42 (100%) <0.001 Data are number (%) or mean ± SD or median (range). Significance was defined as P < 0.05. *Liver histology was reviewed in 169 liver specimens. †Inflammation grade: 0, normal; I, minimal; II, mild; III, moderate; and IV, severe. ‡Fibrosis stage: 0, normal; I, portal fibrosis; II, periportal fibrosis; III, septal fibrosis; and IV, cirrhosis.

Variables HVPG < 10 mm Hg (n = 129) HVPG ≥ 10 mm Hg (n = 42) Univariate Analysis, P Multivariate Analysis, P Age (years) 49 (16–84) 49 (36–69) 0.281 — Male sex 88 (68.75%) 31 (72.1%) 0.551 — Weight (kg) 63.62 ± 9.2 66.98 ± 9.5 0.118 — Height (cm) 165.66 ± 7.5 166.14 ± 8.7 0.767 — BMI (kg/m2) 23.1 (17.1–33.1) 24.1 (19.7–32.3) 0.068 — ICG-R15% 13.93 ± 13.9 39.11 ± 15.0 <0.001 <0.001 Platelet count (×1000) 241.18 ± 80.7 91.89 ± 72.4 <0.001 0.003 PT-INR 1.06 ± 0.1 1.83 ± 0.7 <0.001 0.014 Serum creatinine (mg/dL) 0.85 ± 0.1 0.83 ± 0.2 0.612 — BUN (mg/dL) 24.61 ± 17.2 26.10 ± 18.0 0.701 — Serum total bilirubin (mg/dL) 0.82 ± 0.3 5.24 ± 8.4 0.003 0.652 Serum AST (U/L) 43.43 ± 26.1 45.67 ± 28.9. 0.367 — Serum ALT (U/L) 45.52 ± 34.2 42.40 ± 41.8 0.277 — Serum γ-GT (U/L) 81.30 ± 34.2 79.8 ± 39.8 0.432 — Serum albumin (g/dL) 4.04 ± 0.4 3.12 ± 0.5 <0.001 0.004 Liver histology* Steatosis >10% 24 (18.9%) 8 (19%) 0.797 Inflammation 0, I, II/III, IV† 122 (96.1%)/5 (3.9%) 37 (88.1%)/5 (11.9%) 0.070 Fibrosis 0, I, II, III/IV‡ 89 (70.1%)/38 (29.9%) 0/42 (100%) <0.001 Data are number (%) or mean ± SD or median (range). Significance was defined as P < 0.05. *Liver histology was reviewed in 169 liver specimens. †Inflammation grade: 0, normal; I, minimal; II, mild; III, moderate; and IV, severe. ‡Fibrosis stage: 0, normal; I, portal fibrosis; II, periportal fibrosis; III, septal fibrosis; and IV, cirrhosis. ALT indicates alanine transaminase; AST, aspartate transaminase; BUN, blood urea nitrogen; γ-GT, gamma-glutamyl transpeptidase; ICG-R15, indocyanine green 15 minutes retention rate; PT-INR, prothrombin time-international normalized ratio.

‡Fibrosis stage: 0, normal; I, portal fibrosis; II, periportal fibrosis; III, septal fibrosis; and IV, cirrhosis. ALT indicates alanine transaminase; AST, aspartate transaminase; BUN, blood urea nitrogen; γ-GT, gamma-glutamyl transpeptidase; ICG-R15, indocyanine green 15 minutes retention rate; PT-INR, prothrombin time-international normalized ratio. TABLE 4 Comparison of Perioperative Outcomes Between Group A (Indocyanine Green 15 Minute Retention Rate <20%) and Group B (Indocyanine Green 15 Minute Retention Rate >20%) in the Validation Cohort

ALT indicates alanine transaminase; AST, aspartate transaminase; BUN, blood urea nitrogen; γ-GT, gamma-glutamyl transpeptidase; ICG-R15, indocyanine green 15 minutes retention rate; PT-INR, prothrombin time-international normalized ratio. TABLE 4 Comparison of Perioperative Outcomes Between Group A (Indocyanine Green 15 Minute Retention Rate <20%) and Group B (Indocyanine Green 15 Minute Retention Rate >20%) in the Validation Cohort Outcomes Group A (n = 357) Group B (n = 68) P Preoperative outcomes Age (years) 53 (26–80) 55 (29–77) 0.820 Male sex 281 (78.7%) 55 (80.8%) 0.382 Weight (kg) 62.2 ± 10.2 64.8 ± 9.5 0.306 Height (cm) 167.54 ± 6.2 168.44 ± 7.3 0.901 BMI (kg/m2) 23.8 (16.5–31.3) 24.1 (17.5–33.4) 0.290 ICG-R15 (%) 12.9 (1.2–19.8) 23.65 (20.1–37.7) <0.001 Platelet count (×1000) 163.58 ± 65.1 140.82 ± 56.0 0.245 PT-INR 1.05 ± 0.1 1.07 ± 0.1 0.360 Serum creatinine (mg/dL) 1.08 ± 0.9 1.19 ± 1.3 0.537 Serum total bilirubin (mg/dL) 0.78 ± 0.4 0.87 ± 0.4 0.255 Serum AST (U/L) 43.85 ± 30.7 50.9 ± 37.7 0.427 Serum ALT (U/L) 40.76 ± 31.7 51.5 ± 57.2 0.188 Serum albumin (g/dL) 4.18 ± 0.3 4.03 ± 0.3 0.563 Operative outcomes Tumor size (cm) 3.5 (0.4–23.0) 3.4 (1.0–19.0) 0.557 Tumor number 1 (1–8) 1 (1–6) 0.454 Operation time (minutes) 173.51 ± 81.9 165.04 ± 75.2 0.844 Intraoperative blood loss (mL) 428 ± 105 397 ± 153 0.341 Major/minor resection* 194/163 32/36 0.088 Resection margin <1 cm 143 (40%) 26 (38.2%) 0.927 TNM stage I/II/III† 242 (67.8%)/74 (20.7%)/30 (11%) 46 (67.6%)/13 (19.1%)/8 (11.8%) 0.754 Postoperative outcomes Peak creatinine (mg/dL) 1.15 ± 1.2 1.09 ± 0.3 0.712 Peak total bilirubin (mg/dL) 2.26 ± 1.4 2.20 ± 0.9 0.764 Peak PT-INR 1.45 ± 0.3 1.44 ± 0.3 0.848 Lowest albumin (g/dL) 3.05 ± 0.4 2.95 ± 0.3 0.178 Lowest platelet count (×1000) 103.50 ± 42.8 101.46 ± 44.6 0.749 Complications Postoperative hemorrhage‡ 9 (2.5%) 1 (1.5%) 0.580 Lung complication§ 40 (11.2%) 8 (11.8%) 0.959 Prolonged ascites 18 (4.4%) 2 (2.9%) 0.428 Biliary complication** 11 (3.1%) 3 (4.4%) 0.607 Encephalopathy 0 0 0.893 Need of dialysis 0 1 (1.5%) 0.321 In-hospital mortality 0 1 (1.5%) 0.321 3-month mortality 0 0 0.893 6-month mortality 5 (1.4%) 1 (1.5%) 0.672 Length of hospital stay (day) 12 (6–83) 13 (8–30) 0.190 Tumor recurrence 146 (40.9%) 31 (45.6%) 0.361 5-year survival rates (%) 76.70% 72.20% 0.372 Follow-up periods (months) 22 (4–60) 23.5 (1–60) 0.655 Data are number (%) or mean ± SD or median (range). Significance was defined as P < 0.05.

ity 5 (1.4%) 1 (1.5%) 0.672 Length of hospital stay (day) 12 (6–83) 13 (8–30) 0.190 Tumor recurrence 146 (40.9%) 31 (45.6%) 0.361 5-year survival rates (%) 76.70% 72.20% 0.372 Follow-up periods (months) 22 (4–60) 23.5 (1–60) 0.655 Data are number (%) or mean ± SD or median (range). Significance was defined as P < 0.05. *Major resection means resection of two or more segments. **Biliary complication means bile leakage or bile duct stricture. †TNM stage from AJCC 7th edition. ‡Postoperative hemorrhage includes wound or intraperitoneal hemorrhage. §Lung complication includes pleural effusion or pneumonia. ALT, alanine transaminase; AST, aspartate transaminase; ICG-R15, indocyanine green 15 minutes retention rate; PT-INR, prothrombin time-international normalized ratio. TABLE 5 Comparison in Type of Surgery Between Group A and Group B Type of Surgery Group A (n = 357) Group B (n = 68) P Major hepatectomy (n = 226) 194 (54.3%) 32 (47.1%) 0.474 Right trisectionectomy 7 (2.0%) 2 (2.9%) Right hemihepatectomy 70 (19.6%) 12 (17.7%) Central bisectionectomy 17 (4.8%) 3 (4.4%) Left hemihepatectomy 31 (8.7%) 5 (7.4%) Right anterior sectionectomy 23 (6.4%) 2 (2.9%) Right posterior sectionectomy 19 (5.3%) 3 (4.4%) Left lateral sectionectomy 27 (7.6%) 5 (7.4%) Minor hepatectomy (n = 199) 163 (45.7%) 36 (53.0%) 0.398 Monosegmentectomy 111 (31.1%) 26 (38.3%) Wedge resection 52 (14.6%) 10 (14.7%)

Laparoscopic surgery for patients with gastric cancer has now been employed for more than 2 decades,1 and its use has steadily increased worldwide. A number of randomized controlled trials (RCT's) with small sample sizes,2–6 interim reports from planned large trials7,8 and several meta-analyses9–15 including retrospective observational studies16–18 have indicated that surgical outcomes of laparoscopic gastrectomy (LG) in terms of factors, such as postoperative complications, patient recovery, blood loss, and the number of harvested lymph nodes, are generally comparable to open gastrectomy (OG) and acceptable. Thus, the surgical safety of LG is now well established. It, however, is generally believed that the issue of oncological outcomes, in terms of relapse-free and overall survival, recurrence rate, and pattern of recurrence will not be settled until results of large-scale randomized trials currently underway are published.7,8

the surgical safety of LG is now well established. It, however, is generally believed that the issue of oncological outcomes, in terms of relapse-free and overall survival, recurrence rate, and pattern of recurrence will not be settled until results of large-scale randomized trials currently underway are published.7,8 Clinical decision making must ideally be based on medical evidence that considers the results of RCTs and meta-analyses. Laparoscopic gastrectomy, however, has already been accepted in many countries. In Japan, the number of patients who undergo LG has increased from 1823 in 2003 to 9168 in 2013, according to a survey conducted by the Japanese Society of Endoscopic Surgery,19 thus representing a 5-fold increase over the last decade. In the real world, progress in new techniques continues to advance without waiting for the results of clinical trials, as exemplified by the increasing use of robotic surgery by many surgeons. How should we reconcile this time lag between daily practice and the availability of randomized evidence?

over the last decade. In the real world, progress in new techniques continues to advance without waiting for the results of clinical trials, as exemplified by the increasing use of robotic surgery by many surgeons. How should we reconcile this time lag between daily practice and the availability of randomized evidence? To fill in these evidence gaps while awaiting the outcomes of clinical trials, some investigators have been shifting emphasis to analytic observational approaches using large-scale databases.20,21 Although this can shed light on simple associations between surgical treatments and outcomes, appropriate adjustment for confounding factors is essential in evaluating the effectiveness of interventions. Retrospective studies using propensity score matching (PSM) represent 1 notable approach for such confounder adjustment.22–25 In most previous research to evaluate the effectiveness of surgical interventions, adjustment for confounding factors using propensity score, however, has been less than satisfactory, because factors used for calculating propensity scores were inadequate (only a small subset of probable confounders available were used) and/or inappropriate (postintervention variables such as histopathological information that could be known only after surgery was used). To estimate propensity scores, all preoperative information influencing surgeons’ clinical judgment of whether LG or OG is indicated should be isolated; otherwise, PSM would not be able to adjust for confounding by indication. We have repeatedly pointed out these concerns about inadequate and inappropriate use of the PSM technique.26–28

sity scores, all preoperative information influencing surgeons’ clinical judgment of whether LG or OG is indicated should be isolated; otherwise, PSM would not be able to adjust for confounding by indication. We have repeatedly pointed out these concerns about inadequate and inappropriate use of the PSM technique.26–28 In view of the current state of evidence surrounding LG, we have organized a consensus meeting involving expert laparoscopic surgeons, epidemiologists, and biostatisticians, and established a large-scale multicenter database to estimate propensity scores more precisely through multidisciplinary discussion. We herein report the details of this historical cohort study, “Laparoscopy versus Open Surgery for Clinical Stage I Gastric Cancer” (LOC-1), with the aim of establishing a more precise estimation of propensity scores and examining the effectiveness and safety of LG over OG, while adjusting as completely as possible for possible confounding by indication.

historical cohort study, “Laparoscopy versus Open Surgery for Clinical Stage I Gastric Cancer” (LOC-1), with the aim of establishing a more precise estimation of propensity scores and examining the effectiveness and safety of LG over OG, while adjusting as completely as possible for possible confounding by indication. METHODS Hypothesis and Cohort Development The aim of this study was to verify our hypothesis that LG is not inferior to OG, in terms of overall survival. We designed the entire study protocol a priori, with consideration of the sample size necessary to evaluate noninferiority. Three Japanese cancer-specialized institutions were selected to participate on the basis of the following criteria: being able to offer both open and laparoscopic surgery to patients with gastric cancer; and having facilities for electronic storage of clinical data, including medical records, images, or laboratory data for all consecutive patients with gastric cancer who underwent gastrectomy from January 2006 through December 2012. All patients at these institutions who met the inclusion criteria below were enrolled. This study was conducted with the approval of the institutional review boards at all the participating hospitals. Cancer staging was based on the seventh Union for International Cancer Control (UICC) TNM classification.29

ember 2012. All patients at these institutions who met the inclusion criteria below were enrolled. This study was conducted with the approval of the institutional review boards at all the participating hospitals. Cancer staging was based on the seventh Union for International Cancer Control (UICC) TNM classification.29 PARTICIPANTS The patients enrolled in this study had histologically confirmed gastric adenocarcinoma, were diagnosed as clinical stage I (T1N0, T2N0, or T1N1), and had undergone gastrectomy, including total, subtotal, proximal, or pylorus-preserving gastrectomy. The exclusion criteria included carcinoma in the gastric stump (after previous gastrectomy), the presence of another primary malignancy, and a history of chemotherapy or chemo-radiotherapy.

ge I (T1N0, T2N0, or T1N1), and had undergone gastrectomy, including total, subtotal, proximal, or pylorus-preserving gastrectomy. The exclusion criteria included carcinoma in the gastric stump (after previous gastrectomy), the presence of another primary malignancy, and a history of chemotherapy or chemo-radiotherapy. Data Collection, Propensity Score Matching, and Sample Size Estimation This observational study was not designed to be a formal noninferiority study because of the expected low incidence rate, but was designed to give a best possible interval estimate of hazard ratio (HR) using PSM based on 30 clinically relevant covariates. When designing this study, it was calculated that at least 592 patients per arm were necessary to show the noninferiority of the LG group to the OG group with 5% margin for proportion, 80% power, and a 2-tailed alpha of 5%. A total of 4235 patients were identified from the institutional databases: 2258 patients underwent open surgery and 1977 patients underwent LG. To optimize the accuracy of the propensity score, our study team clarified through their consensus meeting the preoperative information related to the choice made by the surgeon as to whether open surgery or laparoscopic surgery would be used. A total of 30 preoperative factors, including details of the patients’ characteristics and tumor findings, were identified (Supplementary Table 1). To collect precise information on patient and tumor characteristics, investigators who were blind to the outcome looked back on the medical records, stored images, and laboratory data for all patients. In all participating hospitals, upper gastrointestinal endoscopy and abdominal computed tomography were performed for every patient with gastric cancer, and barium swallow or endoscopic ultrasound were available and used for some cases. There was no performance of diagnostic laparoscopy for clinical stage I gastric cancer. Propensity score estimates and matching were derived by 2 biostatisticians (TA and MI) who were also blind to the outcome. The score was estimated using a logistic regression model and greedy matching (ratio = 1:1 without replacement) with a caliper of width 0.2 standard deviations of the logit of the estimated propensity score.

score estimates and matching were derived by 2 biostatisticians (TA and MI) who were also blind to the outcome. The score was estimated using a logistic regression model and greedy matching (ratio = 1:1 without replacement) with a caliper of width 0.2 standard deviations of the logit of the estimated propensity score. In addition to the PSM, 5 factors (clinical T and N factor, esophageal invasion, duodenum invasion, and tumor location as a preoperative diagnosis) were exactly matched to achieve better balance.30 We did not use statistical imputation for missing data because this was an issue in only 23 cases (0.53%). After matching, 924 patients each in the OG and LG groups were included in the final analyses. The balance of each covariate before and after the matching between the 2 groups was evaluated by standardized differences.31 Absolute value of standardized differences less than 10% was considered to be a relatively small imbalance. A flowchart of patient enrollment is shown in Figure 1. FIGURE 1 Study Population. Flowchart of patient enrollment. After matching, 924 patients in both groups were included in the final analyses.

In addition to the PSM, 5 factors (clinical T and N factor, esophageal invasion, duodenum invasion, and tumor location as a preoperative diagnosis) were exactly matched to achieve better balance.30 We did not use statistical imputation for missing data because this was an issue in only 23 cases (0.53%). After matching, 924 patients each in the OG and LG groups were included in the final analyses. The balance of each covariate before and after the matching between the 2 groups was evaluated by standardized differences.31 Absolute value of standardized differences less than 10% was considered to be a relatively small imbalance. A flowchart of patient enrollment is shown in Figure 1. FIGURE 1 Study Population. Flowchart of patient enrollment. After matching, 924 patients in both groups were included in the final analyses. TREATMENT METHODS All institutions that participated in this study were specialized cancer hospitals and all enrolled patients received relatively homogeneous treatments, according to gastric cancer treatment guideline in Japan.32 Although there were no surgeon-specific criteria in this study, considering the average number of patients undergoing gastrectomy was more 300 cases per year in each hospital during this period, all surgeons were considered to have enough experiences to perform both laparoscopic and open surgery. In Japanese specialized cancer hospitals, D2 lymphadenectomy,33 which includes dissection of the region around the splenic artery, celiac artery and hepatic artery without distal pancreatectomy, is considered standard. In this regard, cases with lymphadenectomy around the proper hepatic artery (No. 12a) were omitted in some instances of clinical stage I disease. Modified D2 lymphadenectomy was defined as “D1 + ” in this study. Adjuvant chemotherapy with S-1 for 1 year was performed for most patients with curative resection and pathological stage II, IIIA, or IIIB tumors.34

my around the proper hepatic artery (No. 12a) were omitted in some instances of clinical stage I disease. Modified D2 lymphadenectomy was defined as “D1 + ” in this study. Adjuvant chemotherapy with S-1 for 1 year was performed for most patients with curative resection and pathological stage II, IIIA, or IIIB tumors.34 Outcomes and Statistical Analysis After matching and fixing the enrolled cases, investigators collected all outcome data. The main outcome was overall survival (OS). Secondary outcomes included relapse-free survival (RFS), disease-specific survival, recurrence pattern, the number of harvested lymph nodes, and the incidence of postoperative complications that were grade 3 or more severe according to the Clavien-Dindo classification.35 The OS, RFS, and disease-specific survival were assessed using the Kaplan-Meier method and compared between the OG and LG groups. The hazard ratios (HR) and 95% confidence intervals (CI) were estimated using the unstratified Cox proportional hazards model as primary analyses.36,37 The stratified Cox model for matched pairs was also fitted as a sensitivity analysis. The descriptive statistics were evaluated in other secondary outcomes, and as necessary, continuous variables were compared using Student t tests and categorical variables by Fisher exact test. All statistical tests were 2-sided, and P values of 0.05 or less were considered to indicate statistical significance. All the analyses were performed with SAS version 9.3 (SAS Institute, Cary, NC).

nd as necessary, continuous variables were compared using Student t tests and categorical variables by Fisher exact test. All statistical tests were 2-sided, and P values of 0.05 or less were considered to indicate statistical significance. All the analyses were performed with SAS version 9.3 (SAS Institute, Cary, NC). Subgroup Analyses To confirm the main outcome for all procedures, disease-free survival was compared for each resection type: total, subtotal, proximal or pylorus-preserving gastrectomy, and D1 + or D2 lymphadenectomy. RESULTS Table 1 shows the patient and tumor characteristics for both pre- and post-PSM. Propensity score distributions are shown in Figure 2. FIGURE 2 Distribution of propensity score before and after matching.

Subgroup Analyses To confirm the main outcome for all procedures, disease-free survival was compared for each resection type: total, subtotal, proximal or pylorus-preserving gastrectomy, and D1 + or D2 lymphadenectomy. RESULTS Table 1 shows the patient and tumor characteristics for both pre- and post-PSM. Propensity score distributions are shown in Figure 2. FIGURE 2 Distribution of propensity score before and after matching. Oncological Outcomes The HR for overall mortality, recurrence, and disease-specific mortality in the LG group compared with OS group were 0.75 (95% CI 0.44–1.27; P = 0.290), 1.01 (95% CI 0.55–1.84; P = 0.981), and 1.38 (95% CI 0.61–3.34; P = 0.411), respectively. Estimated HR with the stratified Cox model gave similar results (not shown). Figure 3 shows the survival curves obtained using the Kaplan-Meier method. The 5-year OS and 3-year RFS were 96.3% (95% CI 95.0–97.6) and 97.4% (95% CI 96.4–98.5) in the OG group, and 97.1% (95% CI 95.9–98.3) and 97.7% (95% CI 96.5–98.8) in the LG group, respectively. The risk differences of OS and RFS in the LG group were −0.82% and −0.32%. The number (proportion) of all death occurrence was 33 (3.57%) in the OG group and 24 (2.60%) in the LG group (P = 0.230) and recurrent cases was 22 (2.38%) in the OG group and 21 (2.27%) in the LG group (P = 1.000), respectively. The most common sites of recurrence were the peritoneum and liver (Table 2). The median observation period was 4.4 years (interquartile range: 3.1–5.2 years). In terms of oncological outcomes, there were no significant differences between groups. The timing of event occurrence was shown in Table 3. Most recurrences occurred within 3 postoperative years. There were no significant differences in oncological outcome among any surgical procedures in subgroup analyses; the 5-year OS in OG group versus LG group were 95.0% (95% CI 92.7–96.6) versus 95.1 (95% CI 92.5–96.8; P = 0.4331) in distal gastrectomy, 90.2% (95% CI 79.3–95.5) versus 95.6% (95% CI 88.4–98.4; P = 0.441) in total gastrectomy, 98.4% (95% CI 95.0–99.5) versus 96.8 (95% CI 93.0–98.6; P = 0.387) in pylorus-preserving gastrectomy, 95.0% (95% CI 85.198.4) versus 84.8% (95% CI 67.6–93.2; P = 0.187) in proximal gastrectomy, 95.6% (95% CI 92.5–97.5) versus 96.0% (95% CI 93.8–97.5; P = 0.5931) in D1 + lymphadenectomy and 95.1% (95% CI 92.8–96.6) versus 92.3% (88.0–95.1; P = 0.2637) in D2 lymphadenectomy.

P = 0.387) in pylorus-preserving gastrectomy, 95.0% (95% CI 85.198.4) versus 84.8% (95% CI 67.6–93.2; P = 0.187) in proximal gastrectomy, 95.6% (95% CI 92.5–97.5) versus 96.0% (95% CI 93.8–97.5; P = 0.5931) in D1 + lymphadenectomy and 95.1% (95% CI 92.8–96.6) versus 92.3% (88.0–95.1; P = 0.2637) in D2 lymphadenectomy. FIGURE 3 Survival curves using the Kaplan-Meier method and the number of at-risk patients in the matched cohort. Overall survival rate. The 5-year overall survival was 95.3% in the open gastrectomy group and 94.8% in the laparoscopic gastrectomy group. Relapse-free survival rate.

P = 0.387) in pylorus-preserving gastrectomy, 95.0% (95% CI 85.198.4) versus 84.8% (95% CI 67.6–93.2; P = 0.187) in proximal gastrectomy, 95.6% (95% CI 92.5–97.5) versus 96.0% (95% CI 93.8–97.5; P = 0.5931) in D1 + lymphadenectomy and 95.1% (95% CI 92.8–96.6) versus 92.3% (88.0–95.1; P = 0.2637) in D2 lymphadenectomy. FIGURE 3 Survival curves using the Kaplan-Meier method and the number of at-risk patients in the matched cohort. Overall survival rate. The 5-year overall survival was 95.3% in the open gastrectomy group and 94.8% in the laparoscopic gastrectomy group. Relapse-free survival rate. Surgical Outcomes Table 2 provides the details of surgical procedures and outcomes. Two intraoperative accidents occurred in the OG group: 1 unplanned splenectomy due to injury to the splenic artery and 1 fatal arrhythmia during surgery. In the LG group, 12 cases had to be converted to open surgery: for control of bleeding in 4 cases, and to secure safe oncological manipulation because of serosal invasion by the tumor in 8 cases. There were no significant differences in terms of surgical outcome, such as the number of harvested lymph nodes, pathological TNM stage, or histological type. Details of the postoperative complications are shown in Table 4. The incidence of postoperative complications more severe than grade III was 5.8% in the OG group and 5.1% in the LG group (P = 0.539). There were more cases of pancreatic fistula in the LG group than in the OG group, whereas the latter group showed a higher incidence of wound infection. With regard to long-term complications, the incidence of small bowel obstruction in the OG group was higher than in the LG group, whereas the latter group showed a higher incidence of internal hernia requiring reoperation. Neither length of postoperative hospitalization nor the proportion of readmission rate differed significantly between the groups. No death within 30 postoperative days occurred in either of the groups.

her than in the LG group, whereas the latter group showed a higher incidence of internal hernia requiring reoperation. Neither length of postoperative hospitalization nor the proportion of readmission rate differed significantly between the groups. No death within 30 postoperative days occurred in either of the groups. DISCUSSION Our results suggest that there are no significant differences in OS, RFS, or site of recurrence between OG and LG, and that there are no differences in short-term postoperative complications. These results were also confirmed for all procedure subgroups, including similar oncological outcomes for laparoscopic total gastrectomy with a D1 + lymphadenectomy in early stage gastric cancer. Given the large sample size and the use of strict propensity score estimation and matching, the results of this study seem to establish that LG is neither oncologically nor surgically inferior to OG for stage I gastric cancer. Our study has many advantages over previous studies addressing the same clinical question using PSM. In the present study, we assembled a team, including expert gastric cancer surgeons, epidemiologists, and biostatisticians, and tried to evaluate the propensity of procedural allocation as precisely as possible. This allowed us to finally identify 30 preoperative factors related to surgical decision making. These clinical data were collected by investigators who were blinded to the outcomes. Thanks to recent advances in electronic medical data storage, which has made it possible to store most clinical data, including unstructured data, such as CT or endoscopic images, without any loss or deterioration of quality, in our study, only 0.2% of data were missing. In the literature, 2 previous studies22,24 have addressed the same research question, but they selected only 3 to 5 covariates, such as body mass index, TNM, or comorbidity to calculate the propensity score. It is clear that surgeons choose the most appropriate surgical approach on the basis of much more information, including for example, patient age, tumor size, location, histological findings of biopsy, history of abdominal surgery, performance status, patient selection, institution, and the year the operation was performed. Surgeons are expected to consider all available information to decide the optimal procedure for each patient. Another weakness of previous studies has been that the propensity score has often been estimated using pathological findings, which is postoperative information.

ution, and the year the operation was performed. Surgeons are expected to consider all available information to decide the optimal procedure for each patient. Another weakness of previous studies has been that the propensity score has often been estimated using pathological findings, which is postoperative information. Any decision regarding surgical approach can only be influenced by preoperative factors, such as clinical TNM stage. Adoption of pathological TNM stage would obviously violate the temporal sequence of cause-effect in propensity estimation. We have discussed the difficulty of propensity estimation for surgical interventions in previous articles.27,28,38

proach can only be influenced by preoperative factors, such as clinical TNM stage. Adoption of pathological TNM stage would obviously violate the temporal sequence of cause-effect in propensity estimation. We have discussed the difficulty of propensity estimation for surgical interventions in previous articles.27,28,38 Our study also has some advantages over conventional RCTs, both in terms of internal and external validity. Through accurate estimation of the propensity scores based on our survey, all of the known confounding factors were much better adjusted for in our study than in an RCT. Therefore, this study can be regarded as an epidemiological attempt to come closer to the truth, encompassing the characteristics of a well-designed observational study, or RCT for which all confounding factors, including unknown ones, are adjusted automatically. Some recent review articles have evaluated the quality of PSM.20,21,39,40 These reviews showed that PSM may sometimes overestimate the efficacy of interventions, while some studies using PSM have led to almost the same conclusions as RCTs. Propensity score matching accuracy seems to vary depending on the research question posed or the type of outcome investigated.20 Because the results of RCTs aimed at addressing the same research question will be published later,7,8 we should be able to assess the validity of our study through comparison with those results.

ensity score matching accuracy seems to vary depending on the research question posed or the type of outcome investigated.20 Because the results of RCTs aimed at addressing the same research question will be published later,7,8 we should be able to assess the validity of our study through comparison with those results. One notable feature of our study methodology is an increase in external validity, even over many RCT's. Because the patients we enrolled were acquired consecutively from all participating institutions, many patients who would not be entered into a clinical trial were included in our analyses, such as those with severe comorbidities, the elderly, or those requiring emergency surgery. Therefore we were able to secure a high degree of external validity. Randomized controlled trails aimed at establishing efficacy of a new intervention often need to sacrifice external validity to enhance their internal validity. As a result, patients actually enrolled into a trial tend to account for only a small proportion of the total candidate patients who would receive the intervention in question in the real world. The same would be true of the quality of the surgical interventions: in randomized surgical intervention trials, it is quite difficult to ensure homogeneity of the quality of surgical interventions, unlike clinical trials of drug therapies. This has been regarded as an inevitable limitation of RCTs of surgical interventions but this limitation would not apply to our retrospective cohort study because we have included all surgeons who use LG, OG, or both in the participating institutions.

of surgical interventions, unlike clinical trials of drug therapies. This has been regarded as an inevitable limitation of RCTs of surgical interventions but this limitation would not apply to our retrospective cohort study because we have included all surgeons who use LG, OG, or both in the participating institutions. Lastly, our study has ethical and pragmatic advantages over RCTs in certain respects. Randomized controlled trails are always hampered by a shortage of registered participants, and they tend to require a long period before oncological follow-up can be completed. Even while waiting for the results, advances in surgical techniques or mechanical devices will have been taking place.41 Therefore there is a possibility that even if trials yield positive results, they may not lead to changes in standard treatment, because surgical techniques may have become more refined in the meantime. Indeed, LG has already been accepted as a standard procedure in many countries, even before publication of the results of clinical trials.19 In this situation, the ethical problem of patients being allocated randomly without consideration of technical or mechanical advances seems to be an important issue that needs to be addressed.

s already been accepted as a standard procedure in many countries, even before publication of the results of clinical trials.19 In this situation, the ethical problem of patients being allocated randomly without consideration of technical or mechanical advances seems to be an important issue that needs to be addressed. There are several important limitations to our study. First, there is no guarantee that all confounding factors were included in our database. It might be possible to overlook unmeasurable or unknown but important factors. In addition, we need to discuss the possibility that our results could be extrapolated to patients with gastric cancer worldwide. The epidemiology and treatment of gastric cancer in other East Asian countries is similar to our present study. In Western countries, particularly in North America, there, however, may be significant disadvantageous factors in surgical treatment. These include higher body mass index (BMI) in patients and a greater proportion of patients with advanced stage and upper third cancer, requiring total gastrectomy. It is important to note that it is not known if the results of our series will be able to be reproduced in patients with higher BMI, such as the many patients with severe obesity (BMI > 35 m2/kg) seen in the Western world. Although higher BMI has been linked to a higher incidence of postoperative complications and to longer operative times in previous reports, this has not been shown to contribute to oncological prognosis.42–45 We hypothesize though that, if similar oncologic principles are followed, patients with severe obesity may have similar results, as laparoscopic surgery has been applied safely and routinely in the Western world in these patients for other indications. Another clinical question is the potential use of LG for more advanced cases such as serosa-invasive tumors or patients with bulky lymph node metastasis. We plan to conduct future research to show outcomes of advanced cases using this same rigorous study design, once more data is available.

ients for other indications. Another clinical question is the potential use of LG for more advanced cases such as serosa-invasive tumors or patients with bulky lymph node metastasis. We plan to conduct future research to show outcomes of advanced cases using this same rigorous study design, once more data is available. In conclusion, while it is known that LG has similar postoperative outcomes to open surgery, this is the first study of its kind to show that long-term oncological outcomes are also similar for these 2 procedures. Methods of advanced PSM, such as those used in our study, should be considered in future work to evaluate surgical interventions with multiple potential confounders or when clinical trials are not feasible. Supplementary Material Supplemental Digital Content Disclosure: Mark A. Healy is supported by NIH T32CA009672–24. TABLE 1 Baseline Characteristics Before and After Propensity Score Matching

In conclusion, while it is known that LG has similar postoperative outcomes to open surgery, this is the first study of its kind to show that long-term oncological outcomes are also similar for these 2 procedures. Methods of advanced PSM, such as those used in our study, should be considered in future work to evaluate surgical interventions with multiple potential confounders or when clinical trials are not feasible. Supplementary Material Supplemental Digital Content Disclosure: Mark A. Healy is supported by NIH T32CA009672–24. TABLE 1 Baseline Characteristics Before and After Propensity Score Matching All Patients (n = 3630) Propensity-Matched Patients (n = 1848) Characteristics Open Surgery (n = 1867) % Laparoscopic Surgery (n = 1763) % Standardized Difference Open surgery (n = 924) % Laparoscopic surgery (n = 924) % Standardized Difference Age Mean 64.1 62.0 18.7 63.2 63.3 −0.7 Sex Male 1245 66.7 1130 64.1 5.4 607 65.7 605 65.5 0.5 Female 622 33.3 633 35.9 −5.4 317 34.3 319 34.5 −0.5 Year 2006 356 19.1 137 7.8 33.6 124 13.4 128 13.9 −1.3 2007 312 16.7 185 10.5 18.2 159 17.2 153 16.6 1.7 2008 289 15.5 201 11.4 12.0 135 14.6 146 15.8 −3.3 2009 331 17.7 248 14.1 10.0 175 18.9 167 18.1 2.2 2010 280 15.0 272 15.4 −1.2 155 16.8 147 15.9 2.3 2011 168 9.0 364 20.6 −33.2 96 10.4 106 11.5 −3.5 2012 131 7.0 356 20.2 −39.2 80 8.7 77 8.3 1.2 ASA-PS 1 698 37.4 884 50.1 −25.9 416 45.0 399 43.2 3.7 2 1051 56.3 827 46.9 18.9 466 50.4 489 52.9 −5.0 3 118 6.3 52 2.9 16.1 42 4.5 36 3.9 3.2 BMI, kg/m2 Mean (SD) 22.7 (3.2) 22.6 (3.2) 4.1 22.7 (3.1) 22.8 (3.2) −5.1 History of abdominal surgery 84 4.5 64 3.6 4.4 39 4.2 41 4.4 −1.1 Combined surgery 5 0.3 2 0.1 3.5 1 0.1 2 0.2 −2.7 Site of lesion Upper 364 19.5 293 16.6 7.5 159 17.2 159 17.2 0.0 Upper to middle 90 4.8 40 2.3 13.8 22 2.4 22 2.4 0.0 Middle 795 42.6 882 50.0 −15.0 447 48.4 447 48.4 0.0 Middle to lower 58 3.1 20 1.1 13.7 12 1.3 12 1.3 0.0 Lower 502 26.9 497 28.2 −2.9 269 29.1 269 29.1 0.0 Entire 47 2.5 22 1.2 9.4 15 1.6 15 1.6 0.0 Esohageal invasion 39 2.1 13 0.7 11.5 7 0.8 7 0.8 0.0 Duodenum invasion 12 0.6 3 0.2 7.4 0 0.0 0 0.0 NA Preoperative ER 136 7.3 257 14.6 −23.5 103 11.1 92 10.0 3.9 Clinical T 1a 293 15.7 497 28.2 −30.5 224 24.2 224 24.2 0.0 1b 1002 53.7 1191 67.6 −28.7 632 68.4 632 68.4 0.0 2 572 30.6 75 4.3 74.1 68 7.4 68 7.4 0.0 Clinical N 0 1758 94.2 1733 98.3 −21.8 904 97.8 904 97.8 0.0 1 109 5.8 30 1.7 21.8 20 2.2 20 2.2 0.0 Size of tumor, mm Mean (SD) 30.7 (15.2) 30.2 (15.6) 3.2 29.2 (14.7) 28.9 (14.5) 1.7 Histological findings of Biopsy specimen Well 872 46.7 645 36.6 20.6 396 42.9 389 42.1 1.5 Poor 640 34.3 786 44.6 −21.2 361 39.1 367 39.7 −1.3 Mixed type 354 19.0 329 18.7 0.8 167 18.1 168 18.2 −0.3 Others 0 0.0 2 0.1 −4.8 0 0.0 0 0.0 NA ASA-PS indicates American Society of Anesthesiologists physical s

28.9 (14.5) 1.7 Histological findings of Biopsy specimen Well 872 46.7 645 36.6 20.6 396 42.9 389 42.1 1.5 Poor 640 34.3 786 44.6 −21.2 361 39.1 367 39.7 −1.3 Mixed type 354 19.0 329 18.7 0.8 167 18.1 168 18.2 −0.3 Others 0 0.0 2 0.1 −4.8 0 0.0 0 0.0 NA ASA-PS indicates American Society of Anesthesiologists physical s tatus; BMI, Body mass index; SD, Standard deviation. TABLE 2 Operating and Pathological Findings Open Surgery (n = 924) (%) Laparoscopic Surgery (n = 924) (%) P Procedure TG 103 11.1 102 11.0 0.797 DG 566 61.3 568 61.5 (B-I/B-II/RY) (467/9/90) (390/0/178) PPG 187 20.2 196 21.2 PG 68 7.4 58 6.3 Lymph node dissection D1+ 346 37.4 656 71.0 0.000 D2 578 62.6 268 29.0 Retrieved lymph nodes Median (range) 33 (5–145) 35 (5–94) — Operating time Mean (SD) 167.7 (47.0) 240.0 (58.2) 0.000 Blood loss Mean (SD) 174.1 (181.2) 60.7 (111.7) 0.000 Switch to open surgery — 12 1.3 — Patological T 1 794 85.9 781 84.5 0.449 2 67 7.3 86 9.3 3 38 4.1 42 4.5 4a 14 1.5 15 1.6 4b 1 0.1 0 0.0 Pathological N 0 810 87.7 793 85.8 0.613 1 76 8.2 96 10.4 2 22 2.4 20 2.2 3a 13 1.4 12 1.3 3b 3 0.3 3 0.3 Pathological M 0 922 99.8 923 99.9 1.000 1 2 0.2 1 0.1 Pathological stage IA 741 80.2 712 77.1 0.680 IB 92 10.0 104 11.3 IIA 45 4.9 60 6.5 IIB 23 2.5 28 3.0 IIIA 9 1.0 10 1.1 IIIB 9 1.0 7 0.8 IIIC 3 0.3 2 0.2 IV 2 0.2 1 0.1 Recurrence site Total occurrences 22 2.4 21 2.3 1.000 Peritneum 8 7 Liver 6 8 Lung 1 1 Bone 2 0 Lymphnode 2 4 Local 2 1 Others 1 0 B-1 indicates Billroth I reconstruction; B-2, Billroth II reconstruction; DG, distal gastrectomy; PG, proximal gastrectomy; PPG, pylorus preserving gastrectomy; RY, Roux-en-Y reconstruction; TG, total gastrectomy.

currences 22 2.4 21 2.3 1.000 Peritneum 8 7 Liver 6 8 Lung 1 1 Bone 2 0 Lymphnode 2 4 Local 2 1 Others 1 0 B-1 indicates Billroth I reconstruction; B-2, Billroth II reconstruction; DG, distal gastrectomy; PG, proximal gastrectomy; PPG, pylorus preserving gastrectomy; RY, Roux-en-Y reconstruction; TG, total gastrectomy. TABLE 3 Timing of Events Occurrence Postoperative Year Event Group 1 2 3 4 5 6 7 Total All death OG 2 10 11 6 1 1 2 33 (%)* 6.06 36.4 69.7 87.9 90.9 93.9 100 LG 1 9 8 4 1 1 0 24 (%)* 4.17 41.7 75.0 91.7 95.8 100 100 Disease specific death OG 0 4 4 0 1 0 0 9 (%)* 0.00 44.4 88.9 88.9 100 100 100 LG 1 6 5 1 0 0 0 13 (%)* 7.69 53.8 92.3 100 100 100 100 Recurrence OG 6 11 3 1 0 1 0 22 (%)* 27.3 77.3 90.9 95.5 95.5 100 100 LG 7 5 4 1 1 2 1 21 (%)* 33.3 57.1 76.2 81.0 85.7 95.2 100 *Cumulative percentage. OG indicates open gastrectomy; LG, laparoscopic gastrectomy. TABLE 4 Postoperative Course

Postoperative Year Event Group 1 2 3 4 5 6 7 Total All death OG 2 10 11 6 1 1 2 33 (%)* 6.06 36.4 69.7 87.9 90.9 93.9 100 LG 1 9 8 4 1 1 0 24 (%)* 4.17 41.7 75.0 91.7 95.8 100 100 Disease specific death OG 0 4 4 0 1 0 0 9 (%)* 0.00 44.4 88.9 88.9 100 100 100 LG 1 6 5 1 0 0 0 13 (%)* 7.69 53.8 92.3 100 100 100 100 Recurrence OG 6 11 3 1 0 1 0 22 (%)* 27.3 77.3 90.9 95.5 95.5 100 100 LG 7 5 4 1 1 2 1 21 (%)* 33.3 57.1 76.2 81.0 85.7 95.2 100 *Cumulative percentage. OG indicates open gastrectomy; LG, laparoscopic gastrectomy. TABLE 4 Postoperative Course Open Surgery (n = 924) % Laparoscopic Surgery (n = 924) % P In-hospital stay, median (range) 11 (4–77) 11 (7–144) — Reoperation 5 0.54 5 0.54 1.000 Operative mortality 0 0.00 0 0.00 — Readmission within 60 postoperative day 59 6.39 66 7.14 0.579 Postoperative complication (> = Grade 3) 54 5.84 47 5.09 0.539 Short-term problems Anastomotic leakage 12 1.30 12 1.30 Intra-abdominal bleeding 0 0.00 1 0.11 Intraluminal bleeding 1 0.11 3 0.32 Pancreatic fistula 8 0.87 12 1.30 Abdominal abscess or fluid collection 4 0.43 6 0.65 Wound infection 5 0.54 1 0.11 Stenosis 5 0.54 8 0.87 Enteroparalysis 1 0.11 1 0.11 Ascites 0 0.00 1 0.11 Pneumonia 2 0.22 3 0.32 Cardiac problem 2 0.22 1 0.11 Neurological problem 0 0.00 0 0.00 Others 2 0.22 0 0.00 Long-term problems Small bowel obstruction 12 1.30 0 0.00 Internal hernia 0 0.00 5 0.54

Perforated diverticulitis of the colon is an uncommon serious abdominal condition, and perforation with purulent peritonitis (Hinchey III)1 is even more uncommon.2 The traditional treatment for this group of patients has been open operation with resection of the inflamed and perforated colon with a stoma, that is, the Hartmann procedure. Considerable morbidity has been reported after the Hartmann procedure3 and many patients will never undergo secondary surgery with reversal of the stoma and restored bowel continuity.4 Less invasive types of surgical treatment have thus been considered.5–8 One such procedure is laparoscopy with abdominal lavage, which in a large prospective case series reported good results.5 However, no randomized trials have yet reported any results. As the published evidence primarily includes retrospective series,7 the need for randomized studies is obvious. The aim of this analysis was to compare short-term results of laparoscopic lavage with the Hartmann procedure within a randomized trial “DIverticulitis—LAparoscopic LAvage vs resection (Hartman procedure) for acute diverticulitis with peritonitis” (DILALA). METHODS Trial Design This trial was designed as a prospective, randomized, controlled trial (1:1) of laparoscopic lavage versus open Hartmann procedure. The protocol has previously been described in detail.9 Patients were included at 9 surgical departments in Sweden and Denmark from February 2010 to February 2014. The reports from this trial follow the CONSORT statement when applicable.10

ed, controlled trial (1:1) of laparoscopic lavage versus open Hartmann procedure. The protocol has previously been described in detail.9 Patients were included at 9 surgical departments in Sweden and Denmark from February 2010 to February 2014. The reports from this trial follow the CONSORT statement when applicable.10 Participants Inclusion of patients was based on radiologic examination of the abdomen showing intra-abdominal fluid or gas and a decision to perform surgery followed by the patient's informed consent. After inclusion, patients were taken to the operating room and the procedures were commenced with a diagnostic laparoscopy. The exclusion criteria were as follows: patients not possible to operate due to concomitant disease or patients participating in another randomized trials in conflict with the protocol and end-points of the DILALA trial. When the diagnostic laparoscopy of the abdomen revealed a diverticulitis Hinchey grade III (purulent peritonitis and an inflamed part of the colon), patients were intraoperatively randomized. Patients with Hinchey grade I to II (no free fluid/pus in the abdomen), Hinchey grade IV (fecal contamination), or other pathology at laparoscopy were not eligible for randomization.

d a diverticulitis Hinchey grade III (purulent peritonitis and an inflamed part of the colon), patients were intraoperatively randomized. Patients with Hinchey grade I to II (no free fluid/pus in the abdomen), Hinchey grade IV (fecal contamination), or other pathology at laparoscopy were not eligible for randomization. Participating hospitals were expected to screen patients for possible inclusion. The screening log was established when a hospital entered the study. All admitted patients with a diagnosis of diverticulitis according to the International Classification of Diagnosis (ICD-10) codes K57.2, K57.3, and K57.8 who underwent surgery at the included hospitals were registered in the screening log. Research personnel assured that the block randomization sequence was followed and that clinical data entered into clinical record forms was correct. No center monitored its own data. Interventions Laparoscopic lavage of all 4 quadrants was performed with saline, 3 L or more, of body temperature, until clear fluid was returned. Open Hartmann procedure was performed through a midline incision. All specimens underwent pathology examination. A passive drain was placed in the pelvis in all patients and left in place for at least 24 hours. Both groups were treated postoperatively according to local routines regarding antibiotic treatment, thrombosis prophylaxis and return to oral feeding. Patients were excluded after randomization because of withdrawn consent; cancer or other diagnoses than diverticulitis at surgery or during follow-up (Fig. 1).

4 hours. Both groups were treated postoperatively according to local routines regarding antibiotic treatment, thrombosis prophylaxis and return to oral feeding. Patients were excluded after randomization because of withdrawn consent; cancer or other diagnoses than diverticulitis at surgery or during follow-up (Fig. 1). FIGURE 1 Flowchart of patients included in the trial. Outcomes Four clinical record forms were filled out by the health care professionals. Baseline patient information was collected at inclusion. Details of the operative procedure as well as the postoperative phase until discharge and follow-up until 6 to 12 weeks were collected. All complications were reviewed in detail and retrospectively classified according to Clavien-Dindo,11 but to reduce the risk of misclassification, grade I and II were combined. Primary outcome of this trial was reoperations within 12 months postoperatively, which will be reported when all patients have reached full follow-up. The present data are on short-term clinical outcomes; morbidity, readmissions, reoperations, and mortality. Additional secondary outcomes are described in the protocol article.9

outcome of this trial was reoperations within 12 months postoperatively, which will be reported when all patients have reached full follow-up. The present data are on short-term clinical outcomes; morbidity, readmissions, reoperations, and mortality. Additional secondary outcomes are described in the protocol article.9 Sample Size The DILALA trial primary end-point was reoperations within 12 months. A reduction of the need for further operations of 10% in the group that has undergone laparoscopic lavage was considered clinically relevant. With a statistical power of 80% and a level of significance at 5%, randomization of 64 patients was required. Given the relatively complicated setup for the trial and that all procedures were emergency surgery, the inclusion was set to 80 randomized patients (40 + 40). Randomization Randomization was stratified per hospital and the patients were randomized in blocks of 10 by a professional statistician not involved in the trial.9 The allocation sequence was concealed from the staff at the participating centers by using sequentially numbered thick opaque sealed envelopes. The envelope was opened perioperatively after the operating surgeon had diagnosed the patient as having a diverticulitis Hinchey grade III. The surgeon on call (often not the local study investigator) was responsible for the perioperative randomization.

ters by using sequentially numbered thick opaque sealed envelopes. The envelope was opened perioperatively after the operating surgeon had diagnosed the patient as having a diverticulitis Hinchey grade III. The surgeon on call (often not the local study investigator) was responsible for the perioperative randomization. Statistical Methods This article explores several secondary end-points in the DILALA trial: morbidity and mortality within 30 days as well as mortality within 90 days. Nonparametric statistics were used, and all results are reported as median with range or percentages in parentheses. Mann-Whitney U analysis, χ2 test, and the Fisher exact test were used where appropriate. Ethical Aspects This trial was approved by the Danish ethical committee (Protocol nr. H-4–2009–088) and the Ethical committee in Gothenburg (EPN Dnr 378–09). The trial was registered at ISRCTN for clinical trials ISRCTN82208287 (http://www.controlled-trials.com/ISRCTN82208287). RESULTS The trial included 139 patients and after diagnostic laparoscopy 83 patients were randomized. After exclusions, there were 39 patients in the laparoscopic lavage group and 36 patients in the open Hartmann's group available for analysis (Fig. 1). The screening log was returned by 6 out of 9 participating hospitals. Demographic data are shown in Table 1, and there were no obvious differences between the groups.

The first successful pancreatoduodenectomy was performed as a 2-stage procedure by Walter Kausch in 1909.1 Later, Allen O. Whipple popularized the procedure by a series of 37 pancreatoduodenectomies during his career.2 Because of high mortality, the operation was nearly abandoned in the 1970s.2 In the 1990s, large retrospective series from specialized centers around the world set a benchmark for operative mortality of below 5%.2 Nevertheless, morbidity remains substantial after pancreatoduodenectomy.3–10 The main contributing factor is postoperative pancreatic fistula (POPF), involving leakage of pancreatic juice from the pancreatic anastomosis, which can lead to severe secondary complications such as intra-abdominal abscesses and erosion bleeding.9,11,12 Data regarding the prevention of POPF by application of somatostatin analogues have been controversial thus far,13,14 but a recent randomized trial strongly suggests that pasireotide successfully reduces POPF rates.15 Numerous attempts at improving pancreatic anastomosis techniques to lower POPF rates have been proposed.2,16,17 The hypothesis of this trial dates back to Walter Kausch, who discussed the possibility of anastomosis of the pancreatic remnant to the jejunum (pancreatojejunostomy, PJ) or the stomach (pancreatogastrostomy, PG) in his 1912 original publication of the first successful pancreatoduodenectomy.1

. After exclusions, there were 39 patients in the laparoscopic lavage group and 36 patients in the open Hartmann's group available for analysis (Fig. 1). The screening log was returned by 6 out of 9 participating hospitals. Demographic data are shown in Table 1, and there were no obvious differences between the groups. The preoperative clinical characteristics of the patients were comparable between the 2 groups as shown in Table 2. There were no violations of the randomization, due to technical or other reasons. The perioperative details are reported in Table 3. Patients in the lavage group had significantly shorter operating time, almost one and a half hour (P < 0.0001). There was a difference in number of patients with a suprapubic urinary catheter (Table 3), and overall 15.8% of the patients in the lavage group did not need a urinary catheter compared to no patients in the Hartmann's group (P = 0.025). Postoperative outcomes are shown in Table 4. Patients operated by laparoscopic lavage spent significantly shorter time (median 4 hours) in the recovery unit compared with the Hartmann's group (median 6 hours; P < 0.05). Laparoscopic lavage resulted in shorter hospital stay (median 6 days) compared with the Hartmann's group (median 9 days; P < 0.05) but had a significantly longer period of abdominal drainage (median 3 vs 2 days; P < 0.05). Mortality within 30 days (3/39 vs 0/36) and 90 days (3/39 vs 4/36) did not differ significantly between the groups.

e resulted in shorter hospital stay (median 6 days) compared with the Hartmann's group (median 9 days; P < 0.05) but had a significantly longer period of abdominal drainage (median 3 vs 2 days; P < 0.05). Mortality within 30 days (3/39 vs 0/36) and 90 days (3/39 vs 4/36) did not differ significantly between the groups. Reoperations and complications classified according to Clavien-Dindo11 are presented in Table 5. There were no statistical differences between the 2 groups and neither did the number of complications per patient differ between the 2 groups. Only 2 patients were readmitted. DISCUSSION The main findings of the present trial were that laparoscopic lavage for perforated diverticulitis with purulent peritonitis (Hinchey III) was feasible and safe. Compared with patients undergoing Hartmann's resection, patients treated with laparoscopic lavage were no different in regard to overall morbidity and short-term mortality. The patients also had shorter duration of surgery and shorter hospital stay after laparoscopic lavage. There are currently 4 ongoing randomized trials9,12–14 of which this study is the first to publish results after complete accrual. The primary endpoint of this trial is number of reoperations within 12 months, a follow-up time not yet reached for all patients. The present article reports short-term data within 30 days after operation and mortality within 90 days.

ls9,12–14 of which this study is the first to publish results after complete accrual. The primary endpoint of this trial is number of reoperations within 12 months, a follow-up time not yet reached for all patients. The present article reports short-term data within 30 days after operation and mortality within 90 days. Patients undergoing laparoscopic lavage had significantly shorter duration of surgery probably reflecting the less extensive surgical procedure, but it may also somewhat be due to the conversion in the Hartmann's resection group. These patients both had a laparoscopic and an open procedure performed. After laparoscopic lavage, there was no need for stoma care training, which in part may explain the shorter hospital stay. This will be part of the estimation of health care costs, which will be analyzed after the 12 months’ follow-up.

Patients undergoing laparoscopic lavage had significantly shorter duration of surgery probably reflecting the less extensive surgical procedure, but it may also somewhat be due to the conversion in the Hartmann's resection group. These patients both had a laparoscopic and an open procedure performed. After laparoscopic lavage, there was no need for stoma care training, which in part may explain the shorter hospital stay. This will be part of the estimation of health care costs, which will be analyzed after the 12 months’ follow-up. Diagnostic laparoscopy was feasible in the majority of the patients, and only 4% of the procedures could not be performed due to technical difficulties or severe intra-abdominal inflammation. In patients randomized to laparoscopic lavage the procedure was feasible in all. We found no differences in overall outcomes such as complications or mortality pointing at laparoscopic lavage as a safe alternative to Hartmann's procedure. Mortality after Hartmann's procedure has been reported to be 5.7% to 24%,2,4 which is comparable to our results of 11.4%. Previous reviews have reported considerably lower mortality rates of 1.4% to 1.7% after laparoscopic lavage.7,15 However, these series may be subject to selection bias underestimating mortality rates and it is possible that our mortality rate after laparoscopic lavage of 7.7% probably more closely reflect the true rate in daily clinical practice. Fewer colonic perforations were reported in the laparoscopic lavage group, which probably is due to that the protocol did not require visualization during laparoscopy to define a Hinchey grade III. The handling of the inflamed colon during an open Hartmann's procedure rather than a more advanced disease in the Hartmann's group would be the explanation of the higher rate of visible perforation.

hich probably is due to that the protocol did not require visualization during laparoscopy to define a Hinchey grade III. The handling of the inflamed colon during an open Hartmann's procedure rather than a more advanced disease in the Hartmann's group would be the explanation of the higher rate of visible perforation. The strengths of this trial were its multicenter randomized design and the fact that the majority of the participating hospitals had screening logs to detect possible selection bias. In our trial, we randomized patients after initial diagnostic laparoscopy to avoid including patients with other diagnoses. The LADIES study also used initial diagnostic laparoscopy before randomization but has stopped accrual for laparoscopic lavage due to safety issues.12,16 The LapLAND as well as the SCANDIV study13,14 randomizes patients before laparoscopy, which may result in the inclusion of patients with other diagnoses. Another important strength in our trial was that all statistical analyses were performed as planned beforehand with no post hoc subgroup analysis. We consider the fact that we had 9 participating hospitals from 2 countries as a strength leading to an increased external validity of the results compared with a single center design. The design of the trial did not require any special training for the surgeons, which also enhances the external validity of our results.

e consider the fact that we had 9 participating hospitals from 2 countries as a strength leading to an increased external validity of the results compared with a single center design. The design of the trial did not require any special training for the surgeons, which also enhances the external validity of our results. One limitation of this trial was the relatively large number of patients potentially possible to include but not enrolled due to various reasons. However, the reasons for noninclusion of possible candidates were such that no obvious selection bias seems to have been present. Rather the reasons were, as expected in a trial dealing with emergency conditions, difficulties regarding logistics. Another limitation was that there was no general agreement upon classification of complications at the time of the design of the study. Later, the Clavien-Dindo classification11 has been widely used but it was not recognized in early 2009 when this protocol was written. This is compensated in part as reoperations and readmissions were reported in the CRF but we also retrospectively classified all complications according to the Claiven-Dindo classification, but of course there is a risk of misinterpretations, especially regarding grade I and II. It may also be mentioned that the analyses were not adjusted for multiple testing, thus low P values should be regarded as interesting findings rather than conclusive evidence.

ations according to the Claiven-Dindo classification, but of course there is a risk of misinterpretations, especially regarding grade I and II. It may also be mentioned that the analyses were not adjusted for multiple testing, thus low P values should be regarded as interesting findings rather than conclusive evidence. Our results may have widespread implications in daily clinical practice when treating patients with complicated diverticulitis. There seems to be an international trend toward not resecting the sigmoid colon even after multiple attacks of diverticulitis, due to a limited risk of perforation after recurrence.17 However, when peritonitis is present, it has until now been the routine to perform Hartmann's procedure. If the long-term results of the present trial together with the results of the other randomized trials support the safety of laparoscopic lavage, then hopefully patients may avoid colon resection and stoma creation and hence some of the well-known short- and long-term complications.3 CONCLUSIONS Laparoscopic lavage in Hinchey III–perforated diverticulitis was feasible and in the short-term as safe as Hartmann's procedure. We suggest that widespread implementation of the technique should await long-term results from the ongoing randomized trials.

Our results may have widespread implications in daily clinical practice when treating patients with complicated diverticulitis. There seems to be an international trend toward not resecting the sigmoid colon even after multiple attacks of diverticulitis, due to a limited risk of perforation after recurrence.17 However, when peritonitis is present, it has until now been the routine to perform Hartmann's procedure. If the long-term results of the present trial together with the results of the other randomized trials support the safety of laparoscopic lavage, then hopefully patients may avoid colon resection and stoma creation and hence some of the well-known short- and long-term complications.3 CONCLUSIONS Laparoscopic lavage in Hinchey III–perforated diverticulitis was feasible and in the short-term as safe as Hartmann's procedure. We suggest that widespread implementation of the technique should await long-term results from the ongoing randomized trials. ACKNOWLEDGMENTS The authors thank the following contributors: Marina Åkerblom Sörensson, MD, Department of Surgery, Karlstad Hospital; Lars Ilum, MD, Department of Surgery, Holbaek Hospital; Lars Borly, MD, Department of Surgery, Odense University Hospital; Nicolaj Stilling, MD, Thomas Buchbjerg MD, Haldora Patricia Kristófersdóttir, MD, Jan Luxhoi, MD, Department of Surgery, Odense University Hospital/Svendborg.

artment of Surgery, Karlstad Hospital; Lars Ilum, MD, Department of Surgery, Holbaek Hospital; Lars Borly, MD, Department of Surgery, Odense University Hospital; Nicolaj Stilling, MD, Thomas Buchbjerg MD, Haldora Patricia Kristófersdóttir, MD, Jan Luxhoi, MD, Department of Surgery, Odense University Hospital/Svendborg. Disclosure: This study has been funded by the following institutions: The Alderbertska research foundation, ALF—the Agreement concerning research and education of doctors, Alice Swenzons foundation, Anna-Lisa and Bror Björnssons foundation, the Swedish Society of Medicine, the FrF foundation, the Göteborg Medical Society, the Sahlgrenska University Hospital Health Technology Assessment Center, Johan & Jacob Söderberg's foundation, Magnus Bergvall's foundation, Ruth and Richard Julin's foundation, Signe and Olof Wallenius’ foundation, The Swedish Research Council 2012-1770, The Health & Medical Care Committee of the Regional Executive Board, and Region Västra Götaland. However, the authors declare no conflicts of interest. TABLE 1 Demography of the Study Population

Disclosure: This study has been funded by the following institutions: The Alderbertska research foundation, ALF—the Agreement concerning research and education of doctors, Alice Swenzons foundation, Anna-Lisa and Bror Björnssons foundation, the Swedish Society of Medicine, the FrF foundation, the Göteborg Medical Society, the Sahlgrenska University Hospital Health Technology Assessment Center, Johan & Jacob Söderberg's foundation, Magnus Bergvall's foundation, Ruth and Richard Julin's foundation, Signe and Olof Wallenius’ foundation, The Swedish Research Council 2012-1770, The Health & Medical Care Committee of the Regional Executive Board, and Region Västra Götaland. However, the authors declare no conflicts of interest. TABLE 1 Demography of the Study Population Laparoscopic Lavage (n = 39) Hartmann's Procedure (n = 36) P Missing Data Lavage/Hartmann Age 62 (18–86) 68 (35–88) 0.124 Sex (women/men) 18/21 21/15 0.292 ASA classification 0.222 2 (5.1%)/3 (8.3%) I 7 (18.9%) 8 (24.2%) II 22 (59.5%) 13 (39.4%) III 8 (21.6%) 10 (30.3%) IV 0 2 (6.1%) BMI 25.6 (21–32) 24.9 (19–36) 0.200 8 (20.5%)/8 (22.2%) Previous diverticulitis 5/39 (12.8%) 5/36 (13.9%) 0.892 Previous abdominal surgery 16/39 (41%) 11/36 (30.6%) 0.345 Diabetes 2/35 (5.7%) 2/30 (6.3%) 0.926 4 (10.2%)/4 (11.1%) Cardiovascular disease 18/38 (47.4%) 15/35 (42.9%) 0.699 1 (2.6%)/1 (2.8%) Immunosuppressants 3/35 (8.6%) 5/32 (15.6%) 0.439 4 (10.2%)/4 (11.1%) TABLE 2 Clinical Characteristics Preoperatively

9%) 0.892 Previous abdominal surgery 16/39 (41%) 11/36 (30.6%) 0.345 Diabetes 2/35 (5.7%) 2/30 (6.3%) 0.926 4 (10.2%)/4 (11.1%) Cardiovascular disease 18/38 (47.4%) 15/35 (42.9%) 0.699 1 (2.6%)/1 (2.8%) Immunosuppressants 3/35 (8.6%) 5/32 (15.6%) 0.439 4 (10.2%)/4 (11.1%) TABLE 2 Clinical Characteristics Preoperatively Laparoscopic Lavage (n = 39) Hartmann's Procedure (n = 36) P Missing Data Lavage/Hartmann Leukocyte count (×109/L) 13.6 (4–25) 13.3 (3–22) 0.953 1 (2.6%)/0 C-reactive protein (mg/L) 218 (3–530) 177.5 (1–460) 0.277 1 (2.6%)/0 Body temperature (Celsius) 37.6 (36–41) 37.9 (37–40) 0.737 4 (10.3%)/2 (5.6%) Abdominal examination 0.479 Soft abdomen and local tenderness or palpable mass 5/39 (12.8%) 5/36 (13.9%) Localized peritonitis 16/39 (41.0%) 12/36 (33.3%) Generalized peritonitis 18/39 (46.2%) 19/36 (52.8%) Decision base for surgery Computed Tomography 38/39 (97.4%) 35/35 (100%) 0.340 0/1 (2.8%) Clinical evaluation 36/38 (94.7%) 33/34 (97.1%) 0.623 1 (2.6%)/2 (5.6%) Time from decision until surgery (hh:mm) 3:05 (1:00–13:38) 2:51 (1:11–8:30) 0.725 TABLE 3 Operative Data

) Generalized peritonitis 18/39 (46.2%) 19/36 (52.8%) Decision base for surgery Computed Tomography 38/39 (97.4%) 35/35 (100%) 0.340 0/1 (2.8%) Clinical evaluation 36/38 (94.7%) 33/34 (97.1%) 0.623 1 (2.6%)/2 (5.6%) Time from decision until surgery (hh:mm) 3:05 (1:00–13:38) 2:51 (1:11–8:30) 0.725 TABLE 3 Operative Data Laparoscopic Lavage (n = 39) Hartmann's Procedure (n = 36) P Missing Data Lavage/Hartmann Duration of surgery (hh:mm) 1:08 (0:28–3:14) 2:34 (0:58–4:26) <0.0001 4 (10.3%)/3 (8.3%) Time between end of surgery and end of anesthesia (hh:mm) 0:19 (0:05–0:42) 0:29 (0:00–1:37) <0.0001 4 (10.3%)/3 (8.3%) Additional surgical procedure* 0 2/36 (5.6%) 0.368 4 (10.3%)/0 Amount of saline used for lavage 0.045 0/8 (22.2%) No lavage 0 4/28 (14.3%) 3 L 23/39 (59%) 19/28 (67.9%) 4–5 L 9/39 (23.1%) 4/28 (14.3%) 6–10 L 3/39 (7.7%) 1/28 (3.6%) >10 L 4/39 (10.3%) 0 Inflammation site 0.295 Sigmoid colon 39/39 (100%) 35/36 (97.2%) Rectum 0 1/36 (2.8%) Visible perforation in the colon 2/38 (5.2%) 18/36 (50%) <0.0001 1 (2.6%)/0 Presence of adhesions 0.710 None 15/39 (38.5%) 16/36 (44.4%) Average 20/39 (51.3%) 18/36 (50%) Severe 4/39 (10.3%) 2/36 (5.6%) Adhesions causing technical difficulties 8/37 (21.6%) 5/35 (14.3%) 0.419 2 (5.1%)/1 (2.8%) Conversions from of randomization 0 0 Description of the Hartmann's procedure Location of the proximal resection margin Ileum† 1/36 (2.8%) Left colon 14/36 (38.9%) Sigmoid colon 21/36 (58.3%) Location of the distal resection margin Sigmoid colon 3/36 (8.3%) Rectosigmoid junction 28/36 (77.8%) Rectum 5/36 (13.9%) Drainage 37/39 (94.9%) 30/36 (83.3%) 0.106 2 (5.1%)/0 Urinary catheter Suprapubic catheter 0/39 (0%) 5/36 (13.9%) 0.016 Transurethral catheter 32/38 (84.2%) 31/36 (86.1%) 0.818 1 (2.6%)/0 *Loop ileostomy and appendectomy.

resection margin Sigmoid colon 3/36 (8.3%) Rectosigmoid junction 28/36 (77.8%) Rectum 5/36 (13.9%) Drainage 37/39 (94.9%) 30/36 (83.3%) 0.106 2 (5.1%)/0 Urinary catheter Suprapubic catheter 0/39 (0%) 5/36 (13.9%) 0.016 Transurethral catheter 32/38 (84.2%) 31/36 (86.1%) 0.818 1 (2.6%)/0 *Loop ileostomy and appendectomy. †Small bowel adherent to the inflamed sigmoid colon. The sigmoid colon was the most proximal resection margin on the colon, but the small bowel was also resected. TABLE 4 Short-term Outcome Data Laparoscopic Lavage (n = 39) Hartmann's Procedure (n = 36) P Missing Data Time in recovery unit, h 4 (1–12) 6 (2–44) 0.045 5 (12.8%)/4 (11.1%) Required intensive care, n 5/39 (12.8%) 4/36 (11.1%) 0.802 Number of days with drainage 3 (0–21) 2 (0–17) 0.021 1 (2.6%)/7 (19.4%) Blood transfusion, n 4/39 (10.3%) 2/36 (5.6%) 0.453 Postoperative hospital stay, d 6 (2–27) 9 (4–36) 0.037 1 (2.6%)/2 (5.6%) Stoma related problems, n Skin irritation around the ostomy N/A 5/36(13.9%) Difficulties learning to care for the stoma N/A 11/36 (30.6%) Reoperation within 30 days, n 5/38 (13.2%) 6/35 (17.1%) 0.634 1 (2.6%)/1 (2.8%) Mortality within 30 days, n 3/39 (7.7%) 0/36 (0%) 0.094 Mortality within 90 days, n 3/39 (7.7%) 4/36 (11.4%) 0.583 Readmission within 30 days, n 0/38 2/35 (5.7%) 0.135 1 (2.6%)/1 (2.8%) More than 1 readmission within 30 days, n 0/38 1/35 (5.7%) 0.294 1 (2.6%)/1 (2.8%) TABLE 5 Details on Complications and Reoperations

) Mortality within 30 days, n 3/39 (7.7%) 0/36 (0%) 0.094 Mortality within 90 days, n 3/39 (7.7%) 4/36 (11.4%) 0.583 Readmission within 30 days, n 0/38 2/35 (5.7%) 0.135 1 (2.6%)/1 (2.8%) More than 1 readmission within 30 days, n 0/38 1/35 (5.7%) 0.294 1 (2.6%)/1 (2.8%) TABLE 5 Details on Complications and Reoperations Laparoscopic Lavage Hartmann's Procedure P Missing Data Complications* Classification according to Clavien-Dindo Grade I and II 17/38 (44.7%) 13/35 (37.1%) 0.510 1 (2.6%)/1 (2.8%) Grade IIIa 3/38 (7.9%) 0/35 (0%) 0.090 1 (2.6%)/1 (2.8%) Grade IIIb 4/38 (10.5%) 2/25 (5.7%) 0.455 1 (2.6%)/1 (2.8%) Grade IVa 3/38 (7.9%) 3/35 (8.6%) 0.916 1 (2.6%)/1 (2.8%) Grade IVb 1/38 (2.6%) 1/35 (2.9%) 0.953 1 (2.6%)/1 (2.8%) Grade V 0 0 1 (2.6%)/1 (2.8%) No. complications per patient 1 7/38 (18.4%) 5/35 (14.3%) 1 (2.6%)/1 (2.8%) 2 8/38 (21.1%) 7/35 (20%) 1 (2.6%)/1 (2.8%) 3 1/38 (2.6%) 1/35 (2.9%) 1 (2.6%)/1 (2.8%) 4 2/38 (5.3%) 1/35 (2.9%) 1 (2.6%)/1 (2.8%) 5 2/38 (5.3%) 0/35 (0%) 1 (2.6%)/1 (2.8%) Reoperation within 30 days, n 5/38 (13.2%) 6/35 (17.1%) 0.634 1 (2.6%)/1 (2.8%) Type of reoperation Intra-abdominal abscess 2 1 Retained drainage (required laparoscopy) 1 0 Perforation of the colon (Hinchey IV) 1 0 Suspicion of bleeding 1 0 Stoma complication 0 2 Persistent peritonitis (no new findings after surgery)† 0 1 Haematoma (and inguinal hernia) 0 1 Sepsis 0 1 *A patient can have more than one complication. Five patients in the laparoscopy group and seven patients in the Hartmann's procedure group had more than one Clavien-Dindo 1 and 2 complication.

ma complication 0 2 Persistent peritonitis (no new findings after surgery)† 0 1 Haematoma (and inguinal hernia) 0 1 Sepsis 0 1 *A patient can have more than one complication. Five patients in the laparoscopy group and seven patients in the Hartmann's procedure group had more than one Clavien-Dindo 1 and 2 complication. †Not considered a complication as the patient did not have any sign of complication during the re-laparotomy

Gastric and esophageal cancers are 2 of the most common malignancies in the world. More than 70% of gastric cancer cases and deaths occur in developing countries, and most occur in Eastern Asia.1 There have been distinct changes in the incidences of cancer at different anatomical subsites of the stomach in Western countries.2–4 Many population-based studies showed that the incidence rates of adenocarcinoma of the esophagus and esophagogastric junction (EGJ) were increased in Western countries and Asian countries, such as Japan and Korea.2–8 Although squamous cell carcinoma of the esophagus is still predominant, more than half of newly diagnosed tumors are adenocarcinoma of distal esophagus and EGJ in Western countries and Japan.2–4,6 Incidence of adenocarcinoma at the upper third of the stomach and EGJ has increased over the past 2 decades, with a rate which has exceeded that at any other site of stomach.2,4 In contrast, incidence of adenocarcinoma at distal stomach has gradually declined over the past 50 years.3 However, investigations on changes of esophagogastric junction adenocarcinoma (EGJA) in China remains sparse.

increased over the past 2 decades, with a rate which has exceeded that at any other site of stomach.2,4 In contrast, incidence of adenocarcinoma at distal stomach has gradually declined over the past 50 years.3 However, investigations on changes of esophagogastric junction adenocarcinoma (EGJA) in China remains sparse. Gastroesophageal reflux disease (GERD) was one of the most common gastrointestinal diseases in Europe and the United States with prevalence rates of 10% to 25% in population-based studies.9,10 Reports from many other regions have also demonstrated an increase in the prevalence of GERD.11,12 Some of those with GERD may develop Barrett esophagus, which can lead to esophageal adenocarcinoma. In addition, the prevalence of GERD was also found to be closely associated with the risk of EGJA in some Asian countries.13 Patients with Barrett esophagus need more intensive screening and surveillance for EGJA. Therefore, it is necessary to understand the changes of GERD and EGJA in China over time. However, there were rare reports that simultaneously assessed the trends of EGJA and GERD in China over time. A study by Zhou et al14 showed that there was an increasing trend of gastric cardia cancer over a 12-year period in a northwest province of China.

to understand the changes of GERD and EGJA in China over time. However, there were rare reports that simultaneously assessed the trends of EGJA and GERD in China over time. A study by Zhou et al14 showed that there was an increasing trend of gastric cardia cancer over a 12-year period in a northwest province of China. With the increasing acceptance of Siewert's definition of EGJA, more and more researchers have studied this kind of clinicopathological entity.15 However, there have been few investigations into the change over time of 3 subtypes of EGJA in China. The volume of surgeries done for gastric and esophageal cancers at our institution is large, and the sources of patients come from the Western China, which may serve as a reference for a large population-based study. In this study, we aimed to evaluate the changes of clinicopathological features and surgical treatment of EGJA on the basis of the Siewert classification16 and the prevalence of GERD among the surgical patients with EGJA in a high-volume Chinese hospital during 1988–2012.

With the increasing acceptance of Siewert's definition of EGJA, more and more researchers have studied this kind of clinicopathological entity.15 However, there have been few investigations into the change over time of 3 subtypes of EGJA in China. The volume of surgeries done for gastric and esophageal cancers at our institution is large, and the sources of patients come from the Western China, which may serve as a reference for a large population-based study. In this study, we aimed to evaluate the changes of clinicopathological features and surgical treatment of EGJA on the basis of the Siewert classification16 and the prevalence of GERD among the surgical patients with EGJA in a high-volume Chinese hospital during 1988–2012. MATERIALS AND METHODS Patients Based on the database of patients with gastric cancer in the Department of Gastrointestinal Surgery and patients with esophageal cancer in the Department of Thoracic Surgery of West China Hospital (between January 1988 and December 2012), 5053 patients underwent surgical resection of gastric and distal esophageal adenocarcinomas. Among the 5053 patients, 1723 (34.1%) were anatomically and pathologically classified as those with EGJA15 (Siewert classification: type I, II, and III tumors) and analyzed in this study. The additional exclusion criteria included upper and middle thoracic esophageal cancer, synchronous multiple gastric and esophageal cancers, other types of malignancies of stomach and esophagus, and gastric stump cancer. Changing trends of EGJA and GERD were analyzed in 5 consecutive time periods: from 1988 to 1992 (period 1), from 1993 to 1997 (period 2), from 1998 to 2002 (period 3), from 2003 to 2007 (period 4), and from 2008 to 2012 (period 5).

cancers, other types of malignancies of stomach and esophagus, and gastric stump cancer. Changing trends of EGJA and GERD were analyzed in 5 consecutive time periods: from 1988 to 1992 (period 1), from 1993 to 1997 (period 2), from 1998 to 2002 (period 3), from 2003 to 2007 (period 4), and from 2008 to 2012 (period 5). Siewert Classification and GERD Siewert and Stein16 proposed a widely approved classification of EGJA in 1998 that divided EGJA into 3 subgroups on the basis of the distance from epicenter to EGJ. A lesion was considered to be a tumor of the EGJ if its center was within 5 cm proximal or distal to the anatomical EGJ, as defined by Siewert et al.15 Tumors with the center located from 5 to 1 cm above the EGJ were designated type I, defined as adenocarcinoma of the distal esophagus. Tumors with the center located 1 cm above to 2 cm below the EGJ were designated type II, representing true carcinoma of the cardia. Tumors centered 2 to 5 cm below the EGJ were designated type III, representing subcardial gastric carcinoma. We recorded the margin of the tumor, the EGJ, and the tumor center. The diagnoses of GERD were referred to the standard reported from America.17

ere designated type II, representing true carcinoma of the cardia. Tumors centered 2 to 5 cm below the EGJ were designated type III, representing subcardial gastric carcinoma. We recorded the margin of the tumor, the EGJ, and the tumor center. The diagnoses of GERD were referred to the standard reported from America.17 Clinicopathological Data Clinicopathological data including demographic parameters, maximal diameter of tumor (cm), Borrmann types, pT stage, pN stage, Tumor-Nodes-Metastasis (TNM), and degree of tumor differentiation (well differentiated, moderate, undifferentiated: poor and signet-ring cell type) were reviewed. For type II and type III tumors, the macroscopic type, pathologic degree, T-stage, N-stage, and TNM stages were classified on the basis of Japanese classification of gastric carcinoma: 3rd English edition.18 The TNM stages of type I tumors were classified according to AJCC cancer staging system for esophageal and esophagogastric junction.19 Surgery-related parameters included resection pattern, surgical radicality, number of lymph nodes dissection, combined organ resection, and the short-term surgical outcomes. The major postoperative complications were defined as disease or disorder occurring as a result of gastric cancer or its resection that required reoperation or other intervention. Operative mortality was any death, regardless of reasons, occurring less than 30 days after surgery in or out of the hospital and more than 30 days after the operation during the same hospitalization.20,21

disorder occurring as a result of gastric cancer or its resection that required reoperation or other intervention. Operative mortality was any death, regardless of reasons, occurring less than 30 days after surgery in or out of the hospital and more than 30 days after the operation during the same hospitalization.20,21 Surgical Approaches Before surgery, all patients in our study had obtained chest radiographs, an abdominal ultrasonogram, or a computed tomographic scan for tumor staging. The Siewert type of cancer was diagnosed preoperatively by upper gastrointestinal barium meal or endoscopic findings, which also provided a reference for the surgical approach. Patients in the study mainly underwent abdominal or thoracic incision with total or subtotal gastrectomy or esophagectomy. For type I tumors, a left transthoracic proximal (PG) or total gastrectomy (TG) and esophagectomy was routinely performed in our department of thoracic surgery. The Ivor Lewis thoracoabdominal approach was selectively used for tumor locating at higher level that required anastomosis in the region of the aortic arch. For type II adenocarcinomas invading the distal of esophagus, transhiatal TG or PG combined with mediastinal lymphadenectomy was preferred.22 Thoracoabdominal incision might be performed for subtotal esophagectomy to guarantee curability, if the frozen section of proximal esophageal cutting edge was positive even after repeating resection of distal esophagus. For type III tumors, transabdominal PG or TG was performed in our department of gastrointestinal surgery. D2 lymphadenectomy was routinely performed, whereas D1/D1+ lymphadenectomy was selectively used. The inferior mediastinal or extended (en bloc) lymph node dissection was performed for patients with esophageal involvement. Intraoperative frozen section was a routine procedure aiming to secure the tumor cells free from the resection margins. For the reconstruction, Roux-en-Y anastomosis was mainly adopted for TG and esophagogastric anastomosis after PG. Combined organ resection was selectively performed to achieve a possible curative resection.

perative frozen section was a routine procedure aiming to secure the tumor cells free from the resection margins. For the reconstruction, Roux-en-Y anastomosis was mainly adopted for TG and esophagogastric anastomosis after PG. Combined organ resection was selectively performed to achieve a possible curative resection. Statistical Analysis Continuous variables were shown as mean ± standard deviation and compared by 1-way analysis of variance test. Changes of EGJA subgroup and the categorical variables were analyzed by χ2 tests. The correlation between the proportion of EGJA and the prevalence of GERD for patients was investigated by linear regression. All statistical analyses were 2-sided and performed at less than 0.05 level of significance using software SPSS (SPSS version 19.0 for Windows, IBM Corp., Armonk, NY). RESULTS Anatomical Distribution From 1988 to 2012, a total of 6612 cases of gastric and distal esophageal cancer were included in our gastric and esophageal database; among them, 1559 patients were excluded because of remnant gastric cancer (n = 132), other malignancy of stomach (n = 223), distal esophageal squamous carcinoma (n = 1075), and multicentric gastric or esophageal malignancy (n = 129), leaving a final study cases of 5053. Among all these patients, 413 were treated in period 1, 518 in period 2, 1104 in period 3, 1038 in period 4, and 1980 in period 5 (Table 1).

malignancy of stomach (n = 223), distal esophageal squamous carcinoma (n = 1075), and multicentric gastric or esophageal malignancy (n = 129), leaving a final study cases of 5053. Among all these patients, 413 were treated in period 1, 518 in period 2, 1104 in period 3, 1038 in period 4, and 1980 in period 5 (Table 1). The proportion of distal esophageal adenocarcinoma was 18.7% among all distal esophageal cancers. Among the 5053 cases, 1723 (34.1%) cases were classified as EGJA, whereas the other 3330 cases were not: tumors of 523 cases were located at the body of the stomach (10.4%), 2605 at the antrum (51.6%), and 202 cases invaded the entire stomach (4.0%). There were just 92 patients with EGJA undergoing surgical resection in the period 1, whereas this number had increased to 707 in the period 5. A gradual increasing in proportion of EGJA was seen: from 22.3% in period 1 to 35.7% in period 5 (P < 0.001). By contrast, a significant decrease was observed in gastric adenocarcinoma of antrum, from 63.7% in period 1 to 50.5% in period 5 (P < 0.001). The proportion of adenocarcinoma in the entire stomach remained relatively stable during the past 25 years (P = 0.324) (Table 1 and Fig. 1) FIGURE 1 Time trends of gastric adenocarcinoma at different anatomical locations over the 25-year period from 1988 to 2012.

have been proposed.2,16,17 The hypothesis of this trial dates back to Walter Kausch, who discussed the possibility of anastomosis of the pancreatic remnant to the jejunum (pancreatojejunostomy, PJ) or the stomach (pancreatogastrostomy, PG) in his 1912 original publication of the first successful pancreatoduodenectomy.1 Almost all retrospective studies suggest superiority of PG over PJ in terms of reduced POPF and other complications.18 To date, however, conflicting results have been reported from 8 prospective randomized controlled trials (RCTs) published from 1995 to 201419–26 (see Supplemental Digital Content Table S1, available at): Only 3 RCTs22,24,25 have demonstrated a reduced rate of POPF after PG, and 4 RCTs20,22,24,25 found advantages of PG over PJ in terms of postoperative complications. Soft pancreatic texture was identified as a risk factor for POPF and other complications in 4 RCTs.19,21,23,25 However, the available RCTs have some limitations. With the exception of the recent Belgian multicenter RCT24 including 329 patients, total case numbers of the RCTs are relatively low (n = 90–151) and only 2 RCTs are multicenter trials. Definitions of perioperative outcomes vary as early trials did not use the current consensus definitions of specific complications in pancreatic surgery established by the International Study Group for Pancreatic Surgery (ISGPS). Although many technical variations of PG and PJ have been reported,16,17 all 8 RCTs were restricted to specific subtypes of PG and PJ. Only 2 RCTs with contradictory results report on postoperative pancreatic function measured during follow-up of 3 to 12 months: the Egyptian trial26 reports worse and the Spanish trial25 reports better pancreatic function. None of the RCTs report on quality of life during follow-up.

By contrast, a significant decrease was observed in gastric adenocarcinoma of antrum, from 63.7% in period 1 to 50.5% in period 5 (P < 0.001). The proportion of adenocarcinoma in the entire stomach remained relatively stable during the past 25 years (P = 0.324) (Table 1 and Fig. 1) FIGURE 1 Time trends of gastric adenocarcinoma at different anatomical locations over the 25-year period from 1988 to 2012. Proportion Changes of EGJA According to Siewert's Classification The changing trends in subtypes of EGJA according to Siewert's classification were shown in Figure 2. Among all the 1723 patients with EGJA (all periods), 679 (39.4%) had type III tumors, 797 (46.3%) had type II tumors, and 247 (14.3%) had type I tumors. The proportion of type II tumors had declined from 55.4% in period 1 to 37.2% in period 5, whereas the proportion of type III tumors rose from 35.9% in period 1 to 47.0% in period 5 (P < 0.001). And the proportion of type I tumors also rose from 8.7% in period 1 to 15.8% in period 5 (P = 0.012). The proportion of type III tumors had increased rapidly during the past 25 years and had exceeded the proportion of type II tumors in the last period (47.0% vs 37.2%). FIGURE 2 Trend in subtype of esophagogastric junctional adenocarcinoma according to Siewert's classification.

Proportion Changes of EGJA According to Siewert's Classification The changing trends in subtypes of EGJA according to Siewert's classification were shown in Figure 2. Among all the 1723 patients with EGJA (all periods), 679 (39.4%) had type III tumors, 797 (46.3%) had type II tumors, and 247 (14.3%) had type I tumors. The proportion of type II tumors had declined from 55.4% in period 1 to 37.2% in period 5, whereas the proportion of type III tumors rose from 35.9% in period 1 to 47.0% in period 5 (P < 0.001). And the proportion of type I tumors also rose from 8.7% in period 1 to 15.8% in period 5 (P = 0.012). The proportion of type III tumors had increased rapidly during the past 25 years and had exceeded the proportion of type II tumors in the last period (47.0% vs 37.2%). FIGURE 2 Trend in subtype of esophagogastric junctional adenocarcinoma according to Siewert's classification. Changes of Prevalence of GERD in Patients With EGJA The changing trends of prevalence of GERD in patients with EGJA are shown in Figure 3. The prevalence of GERD was 6.5% in period 1 and had increased gradually to 10.9% in period 5 for the 3 subgroups without significant difference (P = 0.459) (Table 2). In patients with Siewert I tumors, the prevalence of GERD fluctuated from 41.3% to 57.1% (P = 0.456). The prevalences of GERD were nearly stable for Siewert II (ranged from 2.4 to 5.9, P = 0.768) and III tumors (ranged from 1.5 to 3.4, P = 0.846), respectively. As regard to the correlation analyses, we analyzed the linear correlation between the proportion of EGJA and the prevalence of GERD for all the patients with EGJA. The results demonstrated that there was an upward tendency with significant difference (r = 0.946, P = 0.000). And there was also a positive correlations with change in GERD for type I tumors (r = 0.438, P = 0.029). However, no such tendencies were observed in type II tumors (r = 0.158, P = 0.449) and type III tumors (r = −0.153, P = 0.466).

d that there was an upward tendency with significant difference (r = 0.946, P = 0.000). And there was also a positive correlations with change in GERD for type I tumors (r = 0.438, P = 0.029). However, no such tendencies were observed in type II tumors (r = 0.158, P = 0.449) and type III tumors (r = −0.153, P = 0.466). FIGURE 3 Time trend of GERD among esophagogastric junctional adenocarcinoma (EGJA) and the proportion trend of GERD among subtype of EGJA according to Siewert's classification. GERD indicates gastroesophageal reflux disease. Changes of Clinicopathological Features for EGJA The general information of patients with EGJA is shown in Table 2. In the 1723 cases with EGJA, the ratio of male to female was 4.4:1, which was much higher than that of overall patients with gastric cancer (2.6:1, data were not shown in the tables). The mean age that was significantly increased from period 1 (55.2 ± 10.0) to period 5 (61.0 ± 9.8) (P < 0.001) for all EGJA cases was (59.5 ± 10.2) years.

with EGJA, the ratio of male to female was 4.4:1, which was much higher than that of overall patients with gastric cancer (2.6:1, data were not shown in the tables). The mean age that was significantly increased from period 1 (55.2 ± 10.0) to period 5 (61.0 ± 9.8) (P < 0.001) for all EGJA cases was (59.5 ± 10.2) years. The clinicopathological features of patients with EGJA are also listed in Table 2. A decreased trend was seen in the maximal tumor size of EGJA from period 1 (6.4 ± 3.0) cm to period 4 (5.3 ± 2.7) cm (P = 0.001). Comparing with period 1, the proportion of Borrmann 2 was increased from 32.6% to 53.3%, whereas the proportion of Borrmann 3 was decreased from 46.7% to 26.6% in period 5 (P < 0.001). The proportion of differentiated tumors decreased from 45.7% in period 1 to 29.8% in period 5, whereas the proportion of undifferentiated tumors increased from 54.3% to 70.2% (P < 0.001). Among the 1723 cases with EGJA, the overall proportion of pT1 tumors was 5.9%. The proportion of pT1 tumors increased gradually with time, from 1.1% in period 1 to 7.5% in period 5 (P = 0.023). Constituent ratios for pTNM stage were shown in Figure 4. The proportion of patients with stage I increased from 4.4% in period 1 to 13.6% in period 5 (P = 0.028), whereas patients with stage III was gradually decreased from 69.6% to 56.0% in period 5 (P = 0.005). The rate of lymph node metastasis was also decreased from 84.8% to 75.1% in period 5 (P = 0.046).

wn in Figure 4. The proportion of patients with stage I increased from 4.4% in period 1 to 13.6% in period 5 (P = 0.028), whereas patients with stage III was gradually decreased from 69.6% to 56.0% in period 5 (P = 0.005). The rate of lymph node metastasis was also decreased from 84.8% to 75.1% in period 5 (P = 0.046). FIGURE 4 Time trends of esophagogastric junctional adenocarcinoma in distribution of TNM stage during the past 25 years in West China Hospital. TNM stage is based on the Japanese classification of gastric carcinoma: 3rd English edition.

d to specific subtypes of PG and PJ. Only 2 RCTs with contradictory results report on postoperative pancreatic function measured during follow-up of 3 to 12 months: the Egyptian trial26 reports worse and the Spanish trial25 reports better pancreatic function. None of the RCTs report on quality of life during follow-up. Here we present data collected at 14 high-volume centers for pancreatic surgery in Germany from the currently largest multicenter randomized trial comparing PG with PJ with respect to perioperative complications and long-term pancreatic function and quality of life.

wn in Figure 4. The proportion of patients with stage I increased from 4.4% in period 1 to 13.6% in period 5 (P = 0.028), whereas patients with stage III was gradually decreased from 69.6% to 56.0% in period 5 (P = 0.005). The rate of lymph node metastasis was also decreased from 84.8% to 75.1% in period 5 (P = 0.046). FIGURE 4 Time trends of esophagogastric junctional adenocarcinoma in distribution of TNM stage during the past 25 years in West China Hospital. TNM stage is based on the Japanese classification of gastric carcinoma: 3rd English edition. Changes of Surgical Treatment and the Postoperative Complications for EGJA Surgery-related information is described in Table 3. Among 1723 patients, the rate of transabdominal TG had increased from 19.6% in period 1 to 42.0% in period 5 (P < 0.001), whereas the rate of PG had declined to 42.7% in period 5 (P < 0.001). Corresponding with these changes, the Roux-en-Y reconstructions had gradually become a dominating mode with the rate increasing to 41.2% in period 5 (P < 0.001). The proportion of R0 resection was also gradually increased from 71.7% to 85.3% during the past 25 years (P = 0.004). The mean number of harvested lymph nodes was significantly increased to 25.2 ± 11.6 in period 5 (P < 0.001). The rate of combined organ resection has dropped from 22.8% to 6.4% in period 5 (P < 0.001). The postoperative complications and postoperative morbidity are presented in Table 4. The rate of patients with postoperative major complications was gradually decreased from 34.8% in period 1 to 17.1% in period 5 (P < 0.001). The overall 30-day mortality rate after resection was 1.0% and did not change with time.

01). The postoperative complications and postoperative morbidity are presented in Table 4. The rate of patients with postoperative major complications was gradually decreased from 34.8% in period 1 to 17.1% in period 5 (P < 0.001). The overall 30-day mortality rate after resection was 1.0% and did not change with time. DISCUSSION In the past 2 decades, we have noticed significant changes in epidemiological characteristics and treatment approaches for EGJA.5–8 However, the incidence of EGJA was rarely reported in China. In our retrospective cohort study, we analyzed the clinicopathological features derived from 5053 cases who underwent surgical resection for gastric and distal esophageal adenocarcinomas during a period of 25 years. In line with previous studies,3–8 we find a significant increasing trend of incidence of EGJA, and the operative method for EGJA has also significantly changed during the past 25 years. Total gastrectomy is becoming a mainstay of surgical treatment of patients with EGJA. The subtype of EGJA had also changed significantly during the past 25 years according to Siewert classification that the proportions of type I and type III tumors were significantly increased whereas type II tumors decreased during the past 25 years. The rate of postoperative complications among patients with EGJA was also significantly decreased during the past 25 years.

ing the past 25 years according to Siewert classification that the proportions of type I and type III tumors were significantly increased whereas type II tumors decreased during the past 25 years. The rate of postoperative complications among patients with EGJA was also significantly decreased during the past 25 years. In this single-institution series, EGJA accounted for 34.1% of all gastric and esophageal adenocarcinomas after surgical resection. Our study showed a significant higher prevalence of EGJA in China than in Western countries.3,4 In consistent with a previous report that indicated a rising trend of gastric cardia cancer in Gansu province of China from 1993 to 2004 by Zhou et al,14 the proportion of EGJA indicated a dramatic rising trend during the past 25 years in our institution. The proportion of EGJA in our institution was substantially higher than available reports from the United States,3,4 Japan,5,6 and Korea.7,8 From our analysis, most of this increase occurred between the first period and the second period. The increasing trend of EGJA may have direct association with the raised detection rate of gastric and esophageal cancers in recent years in China because of the increased applications of various diagnostic modalities such as endoscopy in symptomatic patients. Furthermore, with the strengthened examination by endoscopy in our country, the proportion of early-stage tumors among patients with EGJA was also significantly increased.

l cancers in recent years in China because of the increased applications of various diagnostic modalities such as endoscopy in symptomatic patients. Furthermore, with the strengthened examination by endoscopy in our country, the proportion of early-stage tumors among patients with EGJA was also significantly increased. The proportion of EGJA was increased in recent years; this might be related to changes in risk factors such as alcohol abuse and the increased incidence of GERD.22,23 The prevalence of GERD has been reported to range from 10% to 48% in Asia, which is slightly lower than that in Western countries and increasing year by year.24–27 Although our results had showed no significant difference, the prevalence of GERD had increased gradually from 6.5% in period 1 to 10.9% in period 5 for the 3 subgroups. Epidemiological studies have consistently suggested that symptom of GERD is the main risk factor for Barrett esophagus and esophageal adenocarcinoma.13 Obesity has also been reported to be associated with development of EGJA at an early age,28 and it is suspected that increased reflux may account for the association.29 It has been reported that the increasing prevalence of GERD among population of Asia might be closely associated with the increasing trend of EGJA in Eastern countries.30,31 And our results also demonstrated that there was a tendency between the proportion of EGJA and the prevalence of GERD for all the patients with EGJA with significant differences. However, cohort studies demonstrated that symptoms of GERD were found monthly in almost 50% of US adults and weekly in nearly 20%, but EGJA is not popular in these patients.32 Therefore, the causality between GERD and EGJA still needs to be researched further.32

all the patients with EGJA with significant differences. However, cohort studies demonstrated that symptoms of GERD were found monthly in almost 50% of US adults and weekly in nearly 20%, but EGJA is not popular in these patients.32 Therefore, the causality between GERD and EGJA still needs to be researched further.32 It should be noted that the prevalences of GERD in patients with Siewert type I tumors in each period accounted nearly 50%, which indicated that there might be closed relationships between GERD and type I tumors. Actually, Lagergren et al33 reported that there was a strong and probably causal relation between gastroesophageal reflux and esophageal adenocarcinoma, including type I EGJA. Wang et al34 also found that the annual detection rate of type I EGJA seemed to be positively correlated with reflux esophagitis in time trend. Although our results found the positive correlations with change in GERD for type I tumors, the correlation was not very strong (correlation coefficient = 0.438). Hence, we consider that evidence is inadequate to show the causality between GERD and type I tumors. Wang et al34 also consider that the similarities in change over time in the rate of detection of EGJA and reflux esophagitis could not easily account for causal relationship.

ry strong (correlation coefficient = 0.438). Hence, we consider that evidence is inadequate to show the causality between GERD and type I tumors. Wang et al34 also consider that the similarities in change over time in the rate of detection of EGJA and reflux esophagitis could not easily account for causal relationship. The widespread application of upper gastrointestinal endoscopy in hospitals of China has made the detection rate of gastric and esophageal cancer increasing. Because of the improvement of our upper gastrointestinal endoscopic technology, the detection rate of tumors located in cardia and fundus has also distinctly increased during the past 25 years. On the contrary, the healthy consciousness of Chinese people has greatly shifted during the past 2 decades. The upper gastrointestinal endoscopic examination is becoming more and more acceptable in our country. We also advocated the cancer-screening propaganda to promote the early diagnosis of upper gastrointestinal tumors in recent years. In accordance with these changes, the rate of pT1 or pN0 tumors and stage I disease among patients with EGJA was also significantly increased, whereas the maximal diameter of tumor and the proportion of stage III tumors showed a decreased trend during the past 25 years.

of upper gastrointestinal tumors in recent years. In accordance with these changes, the rate of pT1 or pN0 tumors and stage I disease among patients with EGJA was also significantly increased, whereas the maximal diameter of tumor and the proportion of stage III tumors showed a decreased trend during the past 25 years. The Siewert's classification for EGJA is now widely accepted. We classified EGJA into 3 subtypes according to Siewert's classification. We found significant higher prevalence of type II and type III tumors and lower rate of type I tumors in China when compared with data from Western countries.35–37 Among all the EGJAs in our series, 14.3% were type I, 46.3% were type II, and 39.4% were type III by reviewing the database of gastric and esophageal cancer of our hospital. These findings are similar to reports from Hasegawa et al38 in Japan and Bai et al39 in China. The increased trend of type I tumors could be explained by the improvement on the rate of detection and changed etiological factors over time. The proportion of Siewert type II tumors was significantly decreased whereas type III increased during the past 25 years in our study. This trend was similar to report from Siewert et al35 in Germany and Chung et al40 in Korea. However, Wang et al34 had indicated that type II and type III tumors just exhibited a fluctuated trend from 2000 to 2009 in China. There are still some controversies on the changing trend of type II and III tumors in China. Therefore, the epidemiological survey for large population is necessary to conduct in the future. From our analysis, type III tumors are gradually becoming a dominant type of EGJA in recent decades.

from 2000 to 2009 in China. There are still some controversies on the changing trend of type II and III tumors in China. Therefore, the epidemiological survey for large population is necessary to conduct in the future. From our analysis, type III tumors are gradually becoming a dominant type of EGJA in recent decades. For the operative procedures, we found that transabdominal TG was gradually becoming more and more common for EGJA in our institution. In Western countries, PG is a traditional procedure to preserve a gastric stump as much as possible for upper-third cancer.41 Proximal gastrectomy has also been generally accepted by most gastrointestinal surgeons for tumors in an early stage providing that a sufficient distal resection margin can be ensured.42 However, Yoo et al43 had found that the recurrence rate of TG group was lower than PG group, because it might have been associated with more radical resection extent. In addition, the rate of postoperative complications was higher in patients who underwent PG since the higher incidence of reflux esophagitis and anastomotic stenosis after PG.44 Proximal gastrectomy with distal esophagectomy via transthoracic esophagectomy approach was common for EGJA for a time in our institution. However, Sasako et al22 reported that an abdominal-transhiatal approach for Siewert II/III tumors achieved a better survival. Therefore, EGJA could be removed safely by TG from an abdominal approach. Transhiatal with mediastinal lymphadenectomy plus TG comprises a more radical resection margin that prevents residual disease at the esophageal margin and allows removal of all the perigastric and peridiaphragmatic hiatus lymph nodes.42 Nowadays, for patients with advanced diseases, many surgeons recommend abdominal-transhiatal approach with TG to be the standard procedure to achieve a more radical effect for Siewert II/III tumors of advanced EGJA.43 These may partly explain the decreased rate of left thoracotomy approach that was selected for type I tumors, and a significant decreased trend was observed in our study.

nd abdominal-transhiatal approach with TG to be the standard procedure to achieve a more radical effect for Siewert II/III tumors of advanced EGJA.43 These may partly explain the decreased rate of left thoracotomy approach that was selected for type I tumors, and a significant decreased trend was observed in our study. Surgical resection plus D2 lymph nodes resection is becoming a standard treatment of gastric cancer. With these changes, the mean number of harvested lymph nodes has significantly increased from 6.1 ± 2.9 in period 1 to 25.2 ± 11.6 in the last period in our institution. The dissection and checkout of lymph nodes had become more and more standard during the past 2 decades. The increase of harvested nodes may be due to the improvements in surgical technique and the improvement in examining lymph nodes. Accompanying these aforementioned changes, our radical degree has also significantly increased during the past 25 years. Although lymph nodal dissection has been reported to increase morbidity,45 the rate of postoperative complications in our institution has fallen from 34.8% in period 1 to 17.1% in period 5. This change may be related to the improvement in surgical skills and postoperative care. Improvements in surgical skills include the surgeon's accumulated experience with large volume of patients, the use of monopolar electrocautery and ultrasound scalpel throughout the procedure, and the adoption of standardized lymph nodes dissection techniques. The advancements in perioperative nutritional support may also be associated with the decrease of postoperative complications. The application of total parenteral nutrition after operation can reduce the risk of postoperative complications among patients with poor nutritional status.46

ymph nodes dissection techniques. The advancements in perioperative nutritional support may also be associated with the decrease of postoperative complications. The application of total parenteral nutrition after operation can reduce the risk of postoperative complications among patients with poor nutritional status.46 One limitation of this study is that it comes from a single institution, so the results may not represent the Chinese population well. However, the volume of surgeries done for gastric and esophageal cancers at our institution is large and the sources of patients come from the area of Western China, which may serve as a reference for a large population-based study. This study strove to describe the changing trend of clinicopathological features, surgical treatment, and proportion of EGJA with the data of patients who underwent resection. Another limitation of our study is that it was confined to operated cases, and patients who underwent nonresection surgery (bypass or biopsy only) were excluded. CONCLUSIONS There seems to be a similar increased trend in the incidence of EGJA in China compared with other regions. The Siewert type I and type III had indicated an increased trend, whereas type II had decreased among past 25 years. The prevalence of GERD among EGJA had also shown a gradually increasing trend. However, the causality between GERD and EGJA still needs further research. Total gastrectomy is becoming a more common procedure for EGJA. Acknowledgments Kai Liu and Kun Yang contributed equally to this study.

CONCLUSIONS There seems to be a similar increased trend in the incidence of EGJA in China compared with other regions. The Siewert type I and type III had indicated an increased trend, whereas type II had decreased among past 25 years. The prevalence of GERD among EGJA had also shown a gradually increasing trend. However, the causality between GERD and EGJA still needs further research. Total gastrectomy is becoming a more common procedure for EGJA. Acknowledgments Kai Liu and Kun Yang contributed equally to this study. The authors thank the substantial work of Volunteer Team of Gastric Cancer Surgery (VOLTGA) West China Hospital, Sichuan University, China, for the establishment of gastric cancer database. The author contributions are as follows:

CONCLUSIONS There seems to be a similar increased trend in the incidence of EGJA in China compared with other regions. The Siewert type I and type III had indicated an increased trend, whereas type II had decreased among past 25 years. The prevalence of GERD among EGJA had also shown a gradually increasing trend. However, the causality between GERD and EGJA still needs further research. Total gastrectomy is becoming a more common procedure for EGJA. Acknowledgments Kai Liu and Kun Yang contributed equally to this study. The authors thank the substantial work of Volunteer Team of Gastric Cancer Surgery (VOLTGA) West China Hospital, Sichuan University, China, for the establishment of gastric cancer database. The author contributions are as follows: Jiankun Hu and Xinzu Chen made substantial contributions to conception and design for this study. Kai Liu and Weihan Zhang acquired and analyzed data and Kai Liu drafted the article. Because this a large sample research in our institution, nearly 5000 cases included in our study. Bo Zhang, Zhixin Chen, Jiaping Chen, Yongfan Zhao, and Zongguang Zhou provided large number of cases and gave many important suggestions for this study. Kun Yang, Xiaolong Chen, Jiankun Hu, and Longqi Chen participated in interpreting critically for important intellectual results. Jiankun Hu and Longqi Chen gave final approval of the version to be published. Kun Yang collected the data of Siewert I tumors and GERD and performed the reanalyses in the process of revision. So Kun Yang completed the revision of important contents and was listed as the co–first author. Professor Longqi Chen kindly provided the data from the database of esophageal cancers in the Department of Thoracic Surgery of West China Hospital and interpreted the results and Professor Longqi Chen gave final approval of the version to be published. Hence, Professor Longqi Chen was also listed as the corresponding author with Professor Jiankun Hu.

ided the data from the database of esophageal cancers in the Department of Thoracic Surgery of West China Hospital and interpreted the results and Professor Longqi Chen gave final approval of the version to be published. Hence, Professor Longqi Chen was also listed as the corresponding author with Professor Jiankun Hu. Disclosure: Supported by grants of the National Natural Science Foundation of China (Nos. 81372344, 81301867) and the New Century Excellent Talents in University support program, Ministry of Education of China (2012SCU-NCET-11–0343). The authors declare no conflicts of interest. TABLE 1 Anatomical Distribution of Gastric Adenocarcinoma in the 25-Year Period Between 1988 and 2012 Tumor Location* Period 1, 1988–1992 (n = 413) Period 2, 1993–1997 (n = 518) Period 3, 1998–2002 (n = 1104) Period 4, 2003–2007 (n = 1038) Period 5, 2008–2012 (n = 1980) P† Esophagogastric junction 92 (22.3%) 175 (33.8%) 388 (35.1%) 361 (34.8%) 707 (35.7%) <0.001 Body 35 (8.5%) 60 (11.6%) 106 (9.6%) 127 (12.2%) 195 (9.9%) 0.108 Antrum 263 (63.7%) 259 (50.0%) 567 (51.4%) 516 (49.7%) 1000 (50.5%) <0.001 Entire stomach 23 (5.6%) 24 (4.6%) 43 (3.9%) 34 (3.3%) 78 (3.9%) 0.324 *The definition of tumor location was according to Japanese classification of gastric carcinoma: 3rd English edition. †Comparisons were performed with χ2 test for categorical variables. TABLE 2 Demographics and Clinicopathological Features of 1723 Patients With Esophagogastric Junctional Adenocarcinoma

Tumor Location* Period 1, 1988–1992 (n = 413) Period 2, 1993–1997 (n = 518) Period 3, 1998–2002 (n = 1104) Period 4, 2003–2007 (n = 1038) Period 5, 2008–2012 (n = 1980) P† Esophagogastric junction 92 (22.3%) 175 (33.8%) 388 (35.1%) 361 (34.8%) 707 (35.7%) <0.001 Body 35 (8.5%) 60 (11.6%) 106 (9.6%) 127 (12.2%) 195 (9.9%) 0.108 Antrum 263 (63.7%) 259 (50.0%) 567 (51.4%) 516 (49.7%) 1000 (50.5%) <0.001 Entire stomach 23 (5.6%) 24 (4.6%) 43 (3.9%) 34 (3.3%) 78 (3.9%) 0.324 *The definition of tumor location was according to Japanese classification of gastric carcinoma: 3rd English edition. †Comparisons were performed with χ2 test for categorical variables. TABLE 2 Demographics and Clinicopathological Features of 1723 Patients With Esophagogastric Junctional Adenocarcinoma Period 1 (n = 92) Period 2 (n = 175) Period 3 (n = 388) Period 4 (n = 361) Period 5 (n = 707) P* Sex 0.085 Male 78(84.8%) 154 (88.0%) 321 (82.7%) 286 (79.2%) 566 (80.1%) Female 14 (145.2%) 21 (12.0%) 67 (17.3%) 75 (20.8) 141 (19.9%) Age, yr <0.001 <40 6 (6.5%) 9 (5.1%) 17 (4.4%) 12(3.3%) 18(2.5%) 40–60 56 (60.9%) 82 (46.9%) 180 (46.4%) 164(45.4%) 281(39.8%) >60 30 (32.6%) 84 (48.0%) 191(49.2%) 185(51.2%) 408(57.7%) Age, yr 55.2 ± 10.0 57.5 ± 9.9 59.1 ± 10.8 59.4 ± 9.9 61.0 ± 9.8 <0.001 GERD* 6 (6.5%) 14 (8.0%) 33 (8.5%) 32 (8.9%) 77 (10.9%) 0.459 Siewert's classification† <0.001 Type I 8(8.7%) 14 (8.0%) 50 (12.9%) 63 (17.5%) 112 (15.8%) 0.012 Type II 51 (55.4%) 102 (58.3%) 212 (54.6%) 169 (46.8%) 263 (37.2%) <0.001 Type III 33 (35.9%) 59 (33.7%) 126 (32.5%) 129 (35.7%) 332(47.0%) <0.001 Maximal tumor size (cm) 6.4 ± 3.0 5.9 ± 2.8 5.5 ± 2.3 5.3 ± 2.7 5.9 ± 3.3 0.001 Macroscopic type Superficial 3(3.3%) 3(1.7%) 14(3.6%) 20 (5.5%) 46 (6.5%) 0.027 Borrmann classification‡ <0.001 Type 1 5(5.4%) 15(8.6%) 24(6.2%) 21 (5.8%) 29 (4.1%) 0.178 Type 2 30(32.6%) 58(33.1%) 161(41.5%) 176 (48.8%) 377 (53.3%) <0.001 Type 3 43(46.7%) 83(47.4%) 155(40.0%) 113 (31.3%) 188 (26.6%) <0.001 Type 4 11 (12.0%) 16(11.1%) 34(8.8%) 31 (8.6%) 67 (9.5%) 0.887 Histological grade§ <0.001 Differentiated 42 (45.7%) 75 (42.9%) 147 (37.9%) 99 (27.4%) 211 (29.8%) Undifferentiated 50 (54.3%) 100 (57.1%) 241 (62.1%) 262 (72.6%) 496 (70.2%) Depth of invasion§ 0.010 T1 1 (1.1%) 4 (2.3%) 22 (5.7%) 21 (5.8%) 53 (7.5%) 0.023 T2 9 (9.8%) 17 (9.7%) 36 (9.3%) 33 (9.1%) 70 (9.9%) 0.995 T3 10 (10.9%) 20 (11.4%) 31 (8.0%) 41 (11.4%) 107 (15.1%) 0.013 T4 72 (78.3%) 134 (76.6%) 299 (77.1%) 266 (73.7%) 477 (67.5%) 0.003 Lymph node metastasis 0.046 Negative 14 (15.2%) 31 (17.7%) 76 (19.6%) 86 (23.8%) 176 (24.9%) Positive 78 (84.8%) 144 (82.3%) 312 (80.4%) 275 (76.2%) 531 (75.1%) TNM stage§ 0.002 I 4 (4.4%) 13 (7.4%) 41 (10.6%) 43 (11.9%) 96 (13.6%) 0.028 II 9 (9.8%) 21 (12.0%) 62 (16.0%) 68 (18.8%) 147 (20.8%) 0.009 III 64 (69.6%) 119 (68.0%) 246 (63.4%) 215 (59.6%) 396 (56.0%) 0.005 IV 15 (16.3%) 22 (12.6%) 39 (10.1%) 35 (9.7%) 68 (9.6%) 0.283 *Continuous variables are reported as mean ± standard deviation.

2 I 4 (4.4%) 13 (7.4%) 41 (10.6%) 43 (11.9%) 96 (13.6%) 0.028 II 9 (9.8%) 21 (12.0%) 62 (16.0%) 68 (18.8%) 147 (20.8%) 0.009 III 64 (69.6%) 119 (68.0%) 246 (63.4%) 215 (59.6%) 396 (56.0%) 0.005 IV 15 (16.3%) 22 (12.6%) 39 (10.1%) 35 (9.7%) 68 (9.6%) 0.283 *Continuous variables are reported as mean ± standard deviation. Comparisons were performed with 1-way analysis of variance test for continuous variables and χ2 test for categorical variables. †Type of esophagogastric junctional adenocarcinoma was according to Siewert's classification. ‡Borrmann classification suits only for advanced gastric cancer (type II and type III tumors). §Histologic grade and TNM stage are based on the Japanese classification of gastric carcinoma: 3rd English edition. GERD indicates gastroesophageal reflux disease. TABLE 3 Surgical Information of 1723 Patients With Esophagogastric Junctional Adenocarcinoma

‡Borrmann classification suits only for advanced gastric cancer (type II and type III tumors). §Histologic grade and TNM stage are based on the Japanese classification of gastric carcinoma: 3rd English edition. GERD indicates gastroesophageal reflux disease. TABLE 3 Surgical Information of 1723 Patients With Esophagogastric Junctional Adenocarcinoma Period 1 (n = 92) Period 2 (n = 175) Period 3 (n = 388) Period 4 (n = 361) Period 5 (n = 707) P* Method of operation <0.001 Transhiatal total gastrectomy 18 (19.6%) 36 (20.6%) 87 (22.4%) 106 (29.4%) 297 (42.0%) <0.001 Transhiatal proximal gastrectomy 67 (72.8%) 99 (56.6%) 193 (49.7%) 155 (42.9%) 302 (42.7%) <0.001 Proximal gastrectomy with distal esophagectomy via left thoracotomy 5 (5.4%) 38 (21.7%) 104 (25.8%) 97 (26.9%) 88 (12.4%) <0.001 Total gastrectomy with distal esophagectomy via left thoracotomy 1 (1.1%) 2 (1.1%) 1 (0.3%) 1 (0.3%) 0 (0%) Ivor-Lewis operation 0 (0%) 0 (0%) 2 (0.5%) 2 (0.6%) 20 (2.8%) Thoracoabdominal esophagectomy and gastrectomy 1 (1.1%) 0 (0%) 1 (0.3%) 0 (0%) 0 (0%) Mode of reconstruction <0.001 EGA 72 (78.3%) 137 (78.3%) 297 (76.5%) 252 (69.8%) 390 (55.2%) Roux-en-Y 19 (20.7%) 37 (21.1%) 82 (21.1%) 101 (28.0%) 291 (41.2%) Roux-en-Y + Pouch 0 (0.0%) 0 (0.0%) 2 (0.5%) 4 (1.1%) 19 (2.7%) JIPA† 1 (1.1%) 1 (0.6%) 7 (1.8%) 4 (1.1%) 7 (1.0%) Mean number of harvested lymph nodes 6.1 ± 2.9 8.1 ± 3.6 10.3 ± 4.6 16.6 ± 9.1 25.2 ± 11.6 <0.001 Combined organ resection‡ 21 (22.8%) 19 (10.9%) 36 (9.3%) 26 (7.2%) 45 (6.4%) <0.001 Radicality 0.004 R0 66 (71.7%) 136 (77.7%) 309 (79.6%) 292 (80.9%) 603 (85.3%) R1/R2 26 (28.3%) 39 (22.3%) 79 (20.4%) 69 (19.1%) 104 (14.7%) *Continuous variables are reported as mean ± standard deviation. Comparisons were performed with analysis of variance for continuous variables and χ2 test for categorical variables.

dicality 0.004 R0 66 (71.7%) 136 (77.7%) 309 (79.6%) 292 (80.9%) 603 (85.3%) R1/R2 26 (28.3%) 39 (22.3%) 79 (20.4%) 69 (19.1%) 104 (14.7%) *Continuous variables are reported as mean ± standard deviation. Comparisons were performed with analysis of variance for continuous variables and χ2 test for categorical variables. †JIPA: Jejunal interposition anastomosis—includes esophagojejunal anastomosis and gastrojejunal anastomosis. ‡Combined organs include esophagus, lung, pericardium, pancreas, spleen, colon, small intestine, gallbladder, and diaphragm. EGA indicates esophagus-gastric anastomosis. TABLE 4 Postoperative Complications of 1723 Patients With Esophagogastric Junctional Adenocarcinoma

†JIPA: Jejunal interposition anastomosis—includes esophagojejunal anastomosis and gastrojejunal anastomosis. ‡Combined organs include esophagus, lung, pericardium, pancreas, spleen, colon, small intestine, gallbladder, and diaphragm. EGA indicates esophagus-gastric anastomosis. TABLE 4 Postoperative Complications of 1723 Patients With Esophagogastric Junctional Adenocarcinoma Period 1 (n = 92) Period 2 (n = 175) Period 3 (n = 388) Period 4 (n = 361) Period 5 (n = 707) P* Postoperative complications 32 (34.8%) 47 (26.9%) 88 (22.7%) 69 (19.1%) 121 (17.1%) <0.001 Postoperative pulmonary complications 8 (8.7%) 13 (7.4%) 28 (7.2%) 18 (5.0%) 29 (4.1%) Pancreatic fistula 1 (1.1%) 2 (1.1%) 3 (0.8%) 1 (0.3%) 4 (0.6%) Pneumothorax 2 (2.2%) 4 (2.3%) 4 (1.0%) 6 (1.7%) 8 (1.1%) Chest infection 1 (1.1%) 1 (0.6%) 2 (0.5%) 5 (1.4%) 4 (0.6%) Pleural effusion 2 (2.2%) 1 (0.6%) 4 (1.0%) 4 (1.1%) 4 (0.6%) Chylothorax 0 (0%) 1 (0.6%) 2 (0.5%) 1 (0.3%) 3 (0.4%) Intra-abdominal infection 3 (3.3%) 4 (2.3%) 9 (2.3%) 5 (1.4%) 17 (2.4%) Intra-abdominal hemorrhage 0 (0%) 0 (0%) 2 (0.5%) 1 (0.3%) 2 (0.3%) Leakage 6 (6.5%) 8 (4.6%) 11 (2.8%) 7 (1.9%) 14 (2.0%) Gastroplegia 1 (1.1%) 0 (0%) 4 (1.0%) 2 (0.6%) 5 (0.7%) Wound infection 7 (7.6%) 9 (5.1%) 18 (4.6%) 14 (3.9%) 27 (3.8%) Others† 1 (1.1%) 4 (2.3%) 1 (0.3%) 5 (1.4%) 4 (0.6%) Operative mortality‡ 1 (1.1%) 1 (0.6%) 3 (0.8%) 5 (1.4%) 8 (1.1%) 0.892 *Comparisons were performed with χ2 test for categorical variables.

(2.0%) Gastroplegia 1 (1.1%) 0 (0%) 4 (1.0%) 2 (0.6%) 5 (0.7%) Wound infection 7 (7.6%) 9 (5.1%) 18 (4.6%) 14 (3.9%) 27 (3.8%) Others† 1 (1.1%) 4 (2.3%) 1 (0.3%) 5 (1.4%) 4 (0.6%) Operative mortality‡ 1 (1.1%) 1 (0.6%) 3 (0.8%) 5 (1.4%) 8 (1.1%) 0.892 *Comparisons were performed with χ2 test for categorical variables. †Others include pyothorax, chest hemorrhage, chylous leakage, pancreatitis, cholecystitis, cardiocerebral events, anesthetic mishap, and venous thrombosis. ‡Operative mortality was any death, regardless of cause, occurring less than 30 days after surgery in or out of the hospital and more than 30 days, during the same hospitalization, after the operation.

The incidence of pancreatic fistula after distal pancreatectomy (DP) remains high at 16% to 35% of cases, and it is associated with a higher incidence of life-threatening complications such as intraabdominal abscess, intraabdominal hemorrhage, or sepsis.1–4 A strategy to decrease pancreatic fistula after DP is urgently required. Stapler closure has recently become a standard technique for pancreatic stump closure; however, the multicenter randomized DISPACT trial found that stapler closure did not significantly reduce the incidence of pancreatic fistula after DP in comparison to hand-sewn closure.5 In addition, we previously reported that stapler closure for transection of a thick pancreas (>12 mm) significantly increased the incidence of pancreatic fistula after DP.6,7 The most appropriate closure technique for the pancreatic stump during DP remains controversial.

la after DP in comparison to hand-sewn closure.5 In addition, we previously reported that stapler closure for transection of a thick pancreas (>12 mm) significantly increased the incidence of pancreatic fistula after DP.6,7 The most appropriate closure technique for the pancreatic stump during DP remains controversial. Several studies have demonstrated that pancreatic fistula in a pancreaticojejunostomy (PJ) of the pancreatic stump does not occur in patients who have undergone DP.8–10 Furthermore, our previous study reported that PJ of the pancreatic stump in DP with en bloc celiac axis resection prevents an extremely high amylase level (>4000 IU/L) in the drainage fluid.11 Several studies have evaluated the association between high drain amylase level and pancreatic fistula.12,13 However, no randomized controlled trials (RCTs) have confirmed the efficacy of PJ of the pancreatic stump during DP. Therefore, a randomized controlled multicenter trial was designed to evaluate whether PJ of the pancreatic stump decreases the incidence of pancreatic fistula after DP compared with stapler closure. METHODS Patients Between June 2011 and March 2014, the RCT was conducted at 6 high-volume surgical centers in Japan. This RCT was approved by the Ethical Committee on Clinical Investigation of each institution and registered in accordance with ClinicalTrials.gov (NCT01384617). This study was conducted in accordance with the Declaration of Helsinki, and written informed consent was preoperatively obtained from all participating patients.

s RCT was approved by the Ethical Committee on Clinical Investigation of each institution and registered in accordance with ClinicalTrials.gov (NCT01384617). This study was conducted in accordance with the Declaration of Helsinki, and written informed consent was preoperatively obtained from all participating patients. Eligible participants were adults 20 years or older who were undergoing DP with or without spleen preservation for disease of the pancreatic body or tail without distant metastasis. Both open and laparoscopic procedures were permitted in the protocol. All patients had to have an Eastern Cooperative Oncology Group performance status of at least 1 and adequate organ function the fulfilled the following criteria: white blood cell 3500/mm3 or more or less than 12,000/mm3, neutrophilic leukocyte 2000/mm3 or more, platelet 100,000/mm3 or more, hemoglobin 9.0 g/dL or more, total bilirubin less than 2.0 mg/dL, aspartate aminotransferase and alanine aminotransferase less than 150 IU/L, and creatinine less than 1.5 mg/dL. Patients with severe comorbidity, such as myocardial infarction, respiratory disorder required oxygen inhalation, liver cirrhosis, hemodialysis, or active duplicative malignant disease affecting adverse events, were excluded from this study.

anine aminotransferase less than 150 IU/L, and creatinine less than 1.5 mg/dL. Patients with severe comorbidity, such as myocardial infarction, respiratory disorder required oxygen inhalation, liver cirrhosis, hemodialysis, or active duplicative malignant disease affecting adverse events, were excluded from this study. Randomization After providing written informed consent, patients undergoing DP were randomly assigned to the stapler closure or PJ group. Randomization was stratified by neoadjuvant therapy such as radiochemotherapy or chemotherapy and institution. A central randomization system for every participating institution was applied, and randomization was done preoperatively.

ent, patients undergoing DP were randomly assigned to the stapler closure or PJ group. Randomization was stratified by neoadjuvant therapy such as radiochemotherapy or chemotherapy and institution. A central randomization system for every participating institution was applied, and randomization was done preoperatively. Transection Line of the Pancreas Transection line of the pancreas was preoperatively planned by assessing the location of the tumor based on a preoperative examination with multidetector-row computed tomography. The final decision for the transection line was made intraoperatively by the surgeon based on the location of the tumor and with the objective of obtaining of a safety margin. The intraoperative ultrasound was usually used to mark transection lines with a safety margin. When the transection line especially for the malignant tumor was located nearer the head of the pancreas, the pancreatic transection was performed at the right side of the portal vein. Transection line was classified as follows: “neck (at the right side of the portal vein),” “body (from the left side of the portal vein to the left of the celiac axis),” and “tail (distal to the left of the celiac axis)”. Moreover, intraoperative frozen section examination of the transection margins was performed routinely. Additional resection of the transection margins was performed, if the results of the intraoperative examination were positive for cancer. Operative Procedure Figure 1 shows a schematic drawing of the 2 procedures, stapler closure and PJ.

Transection Line of the Pancreas Transection line of the pancreas was preoperatively planned by assessing the location of the tumor based on a preoperative examination with multidetector-row computed tomography. The final decision for the transection line was made intraoperatively by the surgeon based on the location of the tumor and with the objective of obtaining of a safety margin. The intraoperative ultrasound was usually used to mark transection lines with a safety margin. When the transection line especially for the malignant tumor was located nearer the head of the pancreas, the pancreatic transection was performed at the right side of the portal vein. Transection line was classified as follows: “neck (at the right side of the portal vein),” “body (from the left side of the portal vein to the left of the celiac axis),” and “tail (distal to the left of the celiac axis)”. Moreover, intraoperative frozen section examination of the transection margins was performed routinely. Additional resection of the transection margins was performed, if the results of the intraoperative examination were positive for cancer. Operative Procedure Figure 1 shows a schematic drawing of the 2 procedures, stapler closure and PJ. FIGURE 1 In the stapler closure group (a), the pancreatic parenchyma was transected using Echelon 60 with a gold cartridge (Ethicon Endo-Surgery, Cincinnati, OH). Echelon 60 with a gold cartridge provides precise and uniform wide compression throughout the entire 60-mm length, with compressible thickness to 1.8 mm, and can attach 2 triple-staggered rows of titanium staples. In the PJ group (b), PJ end-to-side anastomosis by a Roux-en-Y limb for the pancreatic stump was performed via a retrocolic route with an appropriate length of the first jejunal loop (at least 30 cm). The anastomosis was performed in a nonstented duct-to-mucosa fashion.

tach 2 triple-staggered rows of titanium staples. In the PJ group (b), PJ end-to-side anastomosis by a Roux-en-Y limb for the pancreatic stump was performed via a retrocolic route with an appropriate length of the first jejunal loop (at least 30 cm). The anastomosis was performed in a nonstented duct-to-mucosa fashion. In the stapler closure group, the pancreatic parenchyma was transected using Echelon 60 with a gold cartridge (Ethicon Endo-Surgery, Cincinnati, OH). Echelon 60 with a gold cartridge provides precise and uniform wide compression throughout the entire 60-mm length, with compressible thickness to 1.8 mm, and can attach 2 triple-staggered rows of titanium staples. In all cases of stapler group, the stapler was operated by manual as follows: the closure jaw was clamped carefully and slowly, taking 10 min at a fixed speed. The stapler was not released immediately after firing, and the jaws of the stapler were held shut for 1 min.

attach 2 triple-staggered rows of titanium staples. In all cases of stapler group, the stapler was operated by manual as follows: the closure jaw was clamped carefully and slowly, taking 10 min at a fixed speed. The stapler was not released immediately after firing, and the jaws of the stapler were held shut for 1 min. In the PJ group,11 the pancreatic parenchyma was resected using an ultrasonic dissector and only a main pancreatic duct was resected, using a scalpel for duct-to-mucosa anastomosis. After resection of the pancreatic parenchyma, PJ end-to-side anastomosis by a Roux-en-Y limb for the pancreatic stump was performed via a retrocolic route with an appropriate length of the first jejunal loop (at least 30 cm). The anastomosis was performed in a nonstented duct-to-mucosa fashion using a single layer of interrupted 5–0 PDS-II® (polydioxanone, Johnson and Johnson Co., Tokyo, Japan). In a seromuscular-parenchymal anastomosis, nonabsorbable interrupted stitches (4–0 Novafil polybutester, Tyco Healthcare Japan Co., Tokyo, Japan) were placed in end-to-side fashion so that the jejunal wall was tightly adherent to the pancreatic stump. A tube stent was not inserted for the duct-to-mucosa anastomosis to avoid having it migrate into the duodenal side.

bsorbable interrupted stitches (4–0 Novafil polybutester, Tyco Healthcare Japan Co., Tokyo, Japan) were placed in end-to-side fashion so that the jejunal wall was tightly adherent to the pancreatic stump. A tube stent was not inserted for the duct-to-mucosa anastomosis to avoid having it migrate into the duodenal side. Postoperative management was identical in both groups. The management of the drains and checking the amylase levels of the drain fluid were standardized in this trial as follows: One 10-mm silicon drain (BLAKE drain, Ethicon Endo-Surgery, NJ) was placed near the stump of the remnant pancreas or anastomosis. The drain was inserted at least until postoperative day (POD) 3. The drain was removed on POD 3 or 4 if the drainage fluid was clear and pancreatic fistula and bacterial contamination were absent. The amylase level in drainage fluid was routinely measured on POD 1, 3, and 4 in all patients with distal pancreatectomy. Prophylactic octreotide to prevent pancreatic fistula was not administrated in this study. All patients received prophylactic antibiotics either intraoperatively only or for 1 or 2 days postoperatively.

amylase level in drainage fluid was routinely measured on POD 1, 3, and 4 in all patients with distal pancreatectomy. Prophylactic octreotide to prevent pancreatic fistula was not administrated in this study. All patients received prophylactic antibiotics either intraoperatively only or for 1 or 2 days postoperatively. Study Endpoints The primary endpoint was the incidence of pancreatic fistula after DP in the 2 randomized groups. Pancreatic fistula was defined based on the guideline from the International Study Group on Pancreatic Fistula (ISGPF), that is, an amylase level in drainage fluid collected on POD 3 that is more than 3 times the serum amylase level.15 Pancreatic fistula was classified as grade A, B, or C according to ISGPF clinical criteria.15 Secondary endpoints were the incidence of clinically relevant pancreatic fistula (ISGPF grade B and C), overall postoperative morbidity, mortality, postoperative hospital stay, and incidence of pancreatic fistula stratified based on the thickness of the pancreatic parenchyma. Postoperative complications such as intra-abdominal abscess, intra-abdominal hemorrhage, and wound infection in this study were grouped according to the Clavien classification.16 Morbidity in this study was redefined as more than grade II on the Clavien classification. Severe complications were defined in this study as a condition that was grade III or more based on the Clavien classification. Delayed gastric emptying (DGE) was defined according to a consensus definition and clinical grading of postoperative DGE proposed by the International Study Group of Pancreatic Surgery (ISGPS).17 DGE was then classified grade A, B, or C by the ISGPS clinical criteria based on the clinical course and postoperative management. The postoperative hospital stay was defined as follows: a return to preoperative activities of daily living, no deep-site infections, normal laboratory data, no drains, and the possibility for oral nutrition above basal metabolism. Mortality was defined as death within 90 days after surgery.

d postoperative management. The postoperative hospital stay was defined as follows: a return to preoperative activities of daily living, no deep-site infections, normal laboratory data, no drains, and the possibility for oral nutrition above basal metabolism. Mortality was defined as death within 90 days after surgery. Data Collection Data were collected prospectively for all patients and included patient demographics, pathologic examination, perioperative clinical information, and complications. Measuring the thickness of the pancreas was follows: the transection line of pancreas was prospectively recorded by the distance from the left edge of the portal vein. The thickness of the transection line of pancreas was estimated and measured by preoperative CT image based on the distance from left portal vein edge measured intraoperatively.

f the pancreas was follows: the transection line of pancreas was prospectively recorded by the distance from the left edge of the portal vein. The thickness of the transection line of pancreas was estimated and measured by preoperative CT image based on the distance from left portal vein edge measured intraoperatively. Statistical Analysis The number of patients required for statistical validity (2-sided test) was based on pancreatic fistula rate. At the design of this protocol, 3 previous studies using ISGPF definition of pancreatic fistula3,4,18 and 2 systematic reviews with meta-analysis19,20 were referred to estimate the incidence of pancreatic fistula for stapler closure. Pancreatic fistula rate after distal pancreatectomy ranges from 20% to 31% in previous literatures. By the result, 25% as pancreatic fistula rate in the stapler closure group was estimated. As pancreatic fistula rate in pancreaticojejunostomy of the pancreatic stump ranges from 0% to 8.6% in other previous literatures,8,9,18 5% was expected as pancreatic fistula rate in the PJ group. Thus, pancreatic fistula rate was estimated to reduce from 25% to 5% by pancreaticojejunostomy of the pancreatic stump. We calculated that this study required 124 patients (62 in each group) to show a difference between the 2 groups at a power of 80% with a significance level of 0.05. Calculating an estimated intraoperative withdrawal rate or postrandomization exclusion of about 10%, it was necessary to enroll a total of 136 patients (68 in each group) to meet the primary endpoint of this study. Furthermore, for intention-to-treat analysis, all randomized patients were analyzed according to the assigned treatment group, except those who did not undergo DP due to peritoneal dissemination or metastasis or because they were switched to another procedure such as pancreaticoduodenectomy or total pancreatectomy. The protocol analysis incorporated only the patients who underwent the assigned procedure.

ording to the assigned treatment group, except those who did not undergo DP due to peritoneal dissemination or metastasis or because they were switched to another procedure such as pancreaticoduodenectomy or total pancreatectomy. The protocol analysis incorporated only the patients who underwent the assigned procedure. Data are expressed as means ± SD or median with range. Patient characteristics and perioperative and postoperative factors between 2 groups were compared by using χ2 statistics, Fisher exact test, and Mann-Whitney U test. Statistical significance was defined as P < 0.05. The statistical analyses were performed using SPSS 20.0 software (SPSS Inc., Chicago, IL).

with range. Patient characteristics and perioperative and postoperative factors between 2 groups were compared by using χ2 statistics, Fisher exact test, and Mann-Whitney U test. Statistical significance was defined as P < 0.05. The statistical analyses were performed using SPSS 20.0 software (SPSS Inc., Chicago, IL). RESULTS During the study period of June 2011 through March 2014, 184 patients were scheduled to undergo DP for benign or malignant disease. A consort flow diagram of this RCT is shown in Figure 2. Of these 184 patients, 48 patients were excluded from the study before randomization for the following reasons: other organ resection such as colon or stomach was required (n = 9), active duplicative malignant disease affecting adverse event was present (n = 8), severe cirrhosis was present (n = 3), neutrophilic leukocyte count was not 2000/mm3 (n = 3) or more, the patient was receiving hemodialysis (n = 2), the patient had a previous gastrectomy (n = 1) or previous jejunostomy (n = 1), and the patient refused to participate (n = 21). The remaining 136 patients were randomly assigned to stapler closure (n = 66) or PJ (n = 70). Five patients in the stapler closure group were subsequently excluded due to 4 peritoneal dissemination (n = 4) or change of procedure (1 total pancreatectomy). In the PJ group, 8 patients were subsequently excluded due to peritoneal dissemination or liver metastasis (n = 6) or change of procedure (1 pancreaticoduodenectomy and 1 central panceratectomy). Four patients who were assigned to the PJ group were switched to stapler closure or hand-sewn suture. Three patients were switched to stapler closure due to an invisible main pancreatic duct at the resection site of the pancreas, and 1 patient was switched to stapler closure due to positive cytology. No patients who were assigned to the stapler closure group were shifted to PJ due to failure of stapler closure of pancreatic stump.

ts were switched to stapler closure due to an invisible main pancreatic duct at the resection site of the pancreas, and 1 patient was switched to stapler closure due to positive cytology. No patients who were assigned to the stapler closure group were shifted to PJ due to failure of stapler closure of pancreatic stump. FIGURE 2 Consort diagram for the trial. Patient Characteristics Table 1 shows the results of histologic analysis of the resected specimens, patient characteristics, preoperative status, and perioperative status. No significant difference was observed with regard to pancreatic cancer (stapler closure, n = 39; PJ, n = 44) and other disease (stapler closure, n = 22; PJ, n = 18) between the 2 groups. No significant differences existed between the 2 groups concerning other patient characteristics. Operative time was significantly longer in the PJ group; 324 min (154–568) compared with the stapler closure group; 219 min (122–626) in median (range) (P < 0.001), although intraoperative bleeding and the rate of transfusion were similar in both groups. Laparoscopic DP was performed in 13.1% of the stapler closure group and 17.7% of the PJ group (P = 0.478). The rate of spleen preservation was similar between the stapler closure group (5.2%) and the PJ group (4.8%) (P = 0.652). TABLE 1 Characteristics of Enrolled Patients

Patient Characteristics Table 1 shows the results of histologic analysis of the resected specimens, patient characteristics, preoperative status, and perioperative status. No significant difference was observed with regard to pancreatic cancer (stapler closure, n = 39; PJ, n = 44) and other disease (stapler closure, n = 22; PJ, n = 18) between the 2 groups. No significant differences existed between the 2 groups concerning other patient characteristics. Operative time was significantly longer in the PJ group; 324 min (154–568) compared with the stapler closure group; 219 min (122–626) in median (range) (P < 0.001), although intraoperative bleeding and the rate of transfusion were similar in both groups. Laparoscopic DP was performed in 13.1% of the stapler closure group and 17.7% of the PJ group (P = 0.478). The rate of spleen preservation was similar between the stapler closure group (5.2%) and the PJ group (4.8%) (P = 0.652). TABLE 1 Characteristics of Enrolled Patients Stapling closure (n = 61) PJ (n = 62) P Age 69 ± 10 66 ± 11 0.093 Sex (male/female) 42/19 37/25 0.289 Body mass index (kg/m2) 22.3 ± 3.0 22.5 ± 3.1 0.684 Diabetes (yes/no) 20/41 14/48 0.206 Preoperative adjuvant therapy (yes/no) 11/50 13/49 0.681 Serum hemoglobin level (g/dL)* 12.9 ± 1.7 12.5 ± 1.7 0.288 Serum creatinine (mg/dL)† 0.89 ± 0.4 0.90 ± 0.3 0.810 Serum albumin level (mg/dL)‡ 4.0 ± 0.4 4.1 ± 0.5 0.358 Serum amylase level (IU/L)§ 104 ± 108 76 ± 35 0.056 Serum C-reactive protein (mg/dL)|| 0.55 ± 1.32 0.34 ± 1.52 0.445 Pancreatic cancer/other disease 39/22 44/18 0.405 Pancreatic cancer 39 39 Intraductal papillary neoplasms 9 7 Neurondocrine tumor 5 4 Mucinous cyst neoplasms 0 3 Serous cyst neoplasms 1 1 Mass-forming pancreatitis 2 1 Other disease 5 2 Operative time, median (range) (min) 229 (122–626) 326 (154–576) <0.001 Intraoperative bleeding, median (range) (ml) 200 (10–3300) 279 (10–1935) 0.083 Red blood cell transfusion (yes/no) 2/59 4/58 0.348 Procedure (open/laparoscopic surgery) 53/8 51/11 0.478 Preservation of spleen (yes/no) 3/58 3/59 0.652 Dissection of SMA plexus (yes/no) 28/33 27/35 0.759 Portal vein resection (yes/no) 4/57 3/59 0.681 Celiac axis resection (yes/no) 4/57 6/56 0.527 Pancreatic texture (soft/hard) 54/7 56/6 0.746 Thickness of the pancreas (mm)¶ 12.8 ± 4.8 13.6 ± 5.0 0.317 *Normal range of hemoglobin level: 12–17.5 g/dL.

o) 3/58 3/59 0.652 Dissection of SMA plexus (yes/no) 28/33 27/35 0.759 Portal vein resection (yes/no) 4/57 3/59 0.681 Celiac axis resection (yes/no) 4/57 6/56 0.527 Pancreatic texture (soft/hard) 54/7 56/6 0.746 Thickness of the pancreas (mm)¶ 12.8 ± 4.8 13.6 ± 5.0 0.317 *Normal range of hemoglobin level: 12–17.5 g/dL. †Normal range of creatinine: 0.53–1.02 mg/dL. ‡Normal range of albumin level: 3.9–4.9 g/dL. §Normal range of amylase level: 15–150 IU/L. ||Normal range of C-reactive protein: 0–0.3 mg/dL. ¶Thickness at the resection site of pancreas, which was measured by preoperative CT images. Postoperative Complications Between Stapler Closure and PJ by Intention-to-treat Analysis In intention-to-treat analysis, the overall incidence of pancreatic fistula occurred in 38.2% (47 of 123 patients) with no significant difference between the 2 group: 37.7% (23 of 61) in stapler closure versus 38.7% (24 of 62) in PJ (P = 0.332) (Table 2). Pancreatic fistula was classified into 3 categories according to ISGPF criteria.15 The proposed clinical grading of pancreatic fistulas in 23 patients in the stapler closure group was grade A (n = 13, 21.3%), grade B (n = 8, 13.1%), and grade C (n = 2, 3.3%). In contrast, the pancreatic fistulas of 24 patients in the PJ group were classified as grade A (n = 18, 29.0%), grade B (n = 6, 9.7%), and grade C (n = 0, 0%). A clinically significant pancreatic fistula (ISGPF classification grade B/C) occurred in 10 patients (16.4%) with stapler closure and 6 patients (9.7%) with PJ (P = 0.201). TABLE 2 Postoperative Complications Based on Intention-to-treat Analysis

Postoperative Complications Between Stapler Closure and PJ by Intention-to-treat Analysis In intention-to-treat analysis, the overall incidence of pancreatic fistula occurred in 38.2% (47 of 123 patients) with no significant difference between the 2 group: 37.7% (23 of 61) in stapler closure versus 38.7% (24 of 62) in PJ (P = 0.332) (Table 2). Pancreatic fistula was classified into 3 categories according to ISGPF criteria.15 The proposed clinical grading of pancreatic fistulas in 23 patients in the stapler closure group was grade A (n = 13, 21.3%), grade B (n = 8, 13.1%), and grade C (n = 2, 3.3%). In contrast, the pancreatic fistulas of 24 patients in the PJ group were classified as grade A (n = 18, 29.0%), grade B (n = 6, 9.7%), and grade C (n = 0, 0%). A clinically significant pancreatic fistula (ISGPF classification grade B/C) occurred in 10 patients (16.4%) with stapler closure and 6 patients (9.7%) with PJ (P = 0.201). TABLE 2 Postoperative Complications Based on Intention-to-treat Analysis Stapling closure (n = 61) PJ (n = 62) P Primary endpoint Pancreatic fistula* 23 (37.7%) 24 (38.7%) 0.332 Grade A 13 (21.3%) 18 (29.0%) Grade B 8 (13.1%) 6 (9.7%) Grade C 2 (3.3%) 0 (0%) Clinically pancreatic fistula†(grade B/C) 10 (16.4%) 6 (9.7%) 0.201 Secondary endpoint Clavien-Dindo classification 30 (49.2%) 35 (56.5%) 0.669 I 13 18 II 9 10 IIIa 7 6 IIIb 0 1 IVa 1 0 IVb 0 0 V 0 0 Severe complication (IIIa or more) 8 (13.1%) 7 (11.3%) 0.757 Intra-abdominal abscess 7 (11.4%) 11 (17.7%) 0.326 Intra-abdominal hemorrhage 2 (3.3%) 0 (0%) 0.261 Delayed gastric emptying (DGE)‡ 5 (8.2%) 4 (6.5%) 0.127 Grade A 5 1 Grade B 0 2 Grade C 0 1 Ileus 3 (3.3%) 1 (1.6%) 0.303 Wound infection 1 (1.6%) 1 (1.6%) 0.748 Reoperation 0 (0%) 1 (1.6%)§ 0.504 Mortality 0 (0%) 0 (0%) 0.999 Percutaneous drainage|| 6 (9.8%) 6 (9.7%) 0.976 Postoperative hospital stay, median (range) (days) 16 (7–98) 16 (8–130) 0.591 *Pancreatic fistula is defined according to the International Study Group of Pancreatic Surgeons (ISGPF) in its pancreatic fistula recommendation.

) 1 (1.6%)§ 0.504 Mortality 0 (0%) 0 (0%) 0.999 Percutaneous drainage|| 6 (9.8%) 6 (9.7%) 0.976 Postoperative hospital stay, median (range) (days) 16 (7–98) 16 (8–130) 0.591 *Pancreatic fistula is defined according to the International Study Group of Pancreatic Surgeons (ISGPF) in its pancreatic fistula recommendation. †Clinical pancreatic fistula is defined as pancreatic fistula grade B/C based on ISGPF. ‡DGE is defined according to ISGPS in its DGE recommendation. §Reoperation due to perforation of the stomach wall, which would be intraoperatively injured by thermal of the device on postoperative day 2. ||Percutaneous drainage undertaken for postoperative management of intra-abdominal abscess related to pancreatic fistula. The overall morbidity rate in this study was 27.6% (34 of 123 patients), with no difference between the 2 group: 27.9% (17 of 61) in the stapler closure group versus 27.4% (17 of 62) in the PJ group (P = 0.956). No significant differences existed between the 2 groups regarding the separate incidence of DGE, intra-abdominal abscess, and intra-abdominal hemorrhage. Although 2 patients in the stapler closure group had intra-abdominal hemorrhage complicated by pancreatic fistula, complete hemostasis could be achieved by interventional radiographic techniques. Intra-abdominal hemorrhage did not occur in any patients in the PJ group. Only 1 patient in the PJ group had reoperation due to perforation of the stomach wall, which would be intraoperatively injured by thermal of the device on POD 2. The mortality rate in this study was zero in both groups.

l radiographic techniques. Intra-abdominal hemorrhage did not occur in any patients in the PJ group. Only 1 patient in the PJ group had reoperation due to perforation of the stomach wall, which would be intraoperatively injured by thermal of the device on POD 2. The mortality rate in this study was zero in both groups. In this study, 19 (15.4%) of 123 patients with distal pancreatectomy underwent laparoscopic distal pancreatectomy. No robotic distal pancreatectomy was performed. Concerning to primary endpoint of the overall incidence of pancreatic fistula, there was no significant difference between the 2 group: 47.3% (9 of 19) in laparoscopic distal pancreatectomy versus 36.5% (38 of 104) in open distal pancreatectomy (P = 0.372). A clinically significant pancreatic fistula (ISGPF classification grade B/C) occurred in 2 patients (10.5%) with laparoscopic distal pancreatectomy and 14 patients (13.5%) with open distal pancreatectomy (P = 0.535). Laparoscopic distal pancreatectomy offered the similar outcomes with open distal pancreatectomy.

.372). A clinically significant pancreatic fistula (ISGPF classification grade B/C) occurred in 2 patients (10.5%) with laparoscopic distal pancreatectomy and 14 patients (13.5%) with open distal pancreatectomy (P = 0.535). Laparoscopic distal pancreatectomy offered the similar outcomes with open distal pancreatectomy. Comparison Between Stapler Closure and PJ in Analysis by Protocol Table 3 shows a comparison between stapler closure and PJ in an analysis by protocol. In this analysis, the overall incidence of pancreatic fistula was 39.5% (47 of 119 patients) with no significant difference between the 2 groups: 37.7% (23 of 61) in the stapler closure group versus 41.4% (24 of 58) in the PJ group (P = 0.356) (Table 3). Pancreatic fistula was classified into 3 categories according to ISGPF criteria.15 The proposed clinical grading of pancreatic fistulas for 23 patients in the stapler closure was grade A (n = 13, 21.3%), grade B (n = 8, 13.1%), and grade C (n = 2, 3.3%). Twenty-four patients in the PJ group had pancreatic fistula classified as grade A (n = 18, 31.1%), grade B (n = 6, 10.3%), or grade C (n = 0, 0%). A clinically significant pancreatic fistula (ISGPF classification Grade B/C) occurred in 10 patients (16.4%) with stapler closure and 6 patients (10.3%) with PJ (P = 0.334). TABLE 3 Pancreatic Fistula and Postoperative Course based on Per-protocol Analysis

Comparison Between Stapler Closure and PJ in Analysis by Protocol Table 3 shows a comparison between stapler closure and PJ in an analysis by protocol. In this analysis, the overall incidence of pancreatic fistula was 39.5% (47 of 119 patients) with no significant difference between the 2 groups: 37.7% (23 of 61) in the stapler closure group versus 41.4% (24 of 58) in the PJ group (P = 0.356) (Table 3). Pancreatic fistula was classified into 3 categories according to ISGPF criteria.15 The proposed clinical grading of pancreatic fistulas for 23 patients in the stapler closure was grade A (n = 13, 21.3%), grade B (n = 8, 13.1%), and grade C (n = 2, 3.3%). Twenty-four patients in the PJ group had pancreatic fistula classified as grade A (n = 18, 31.1%), grade B (n = 6, 10.3%), or grade C (n = 0, 0%). A clinically significant pancreatic fistula (ISGPF classification Grade B/C) occurred in 10 patients (16.4%) with stapler closure and 6 patients (10.3%) with PJ (P = 0.334). TABLE 3 Pancreatic Fistula and Postoperative Course based on Per-protocol Analysis Stapling closure (n = 61) PJ (n = 58) P Pancreatic fistula* 23 (37.7%) 24 (41.4%) 0.356 Grade A 13 (21.3%) 18 (31.1%) Grade B 8 (13.1%) 6 (10.3%) Grade C 2 (3.3%) 0 (0%) Clinically pancreatic fistula† (grade B/C) 10 (16.4%) 6 (10.3%) 0.334 Amylase level of drainage fluid on POD 1, median (range) (IU/L) 1,657 (92–37,410) 938 (122–10,660) 0.244 Amylase level of drainage fluid on POD 3, median (range) (IU/L) 245 (12–96,687) 296 (18–3,454) 0.722 Amylase level of drainage fluid on POD 4, median (range) (IU/L) 125 (7–48,483) 107 (19–3,527) 0.606 The day until first flatus, median (range) (days) 3 (1–19) 3 (1–5) 0.686 Start of solid diet, median (range) (days) 5 (2–22) 4 (2–20) 0.816 Time to drain removal, median (range) (days) 4 (3–83) 4 (3–36) 0.268 Percutaneous drainage‡ 6 (9.8%) 6 (10.3%) 0.927 Reoperation 0 (0%) 0 (0%) 0.999 Postoperative hospital stay, median (range) (days) 16 (7–98) 16 (8–130) 0.727 *Pancreatic fistula is defined according to the International Study Group of Pancreatic Surgeons (ISGPF) in its pancreatic fistula recommendation.

4 (3–36) 0.268 Percutaneous drainage‡ 6 (9.8%) 6 (10.3%) 0.927 Reoperation 0 (0%) 0 (0%) 0.999 Postoperative hospital stay, median (range) (days) 16 (7–98) 16 (8–130) 0.727 *Pancreatic fistula is defined according to the International Study Group of Pancreatic Surgeons (ISGPF) in its pancreatic fistula recommendation. †Clinical pancreatic fistula is defined as pancreatic fistula grade B/C based on ISGPF. ‡Percutaneous drainage undertaken for postoperative management of intra-abdominal abscess related to pancreatic fistula. The amylase level of the drainage fluid on POD 1: 1657 (92–37,410) versus 938 (122–10,660) IU/L, POD 3: 245 (12–96,687) IU/L versus 296 (18–3,454) IU/L, and POD 4: 125 (7–48,483) IU/L versus 107 (19–3,527) IU/L were similar between the stapler closure and PJ groups (P = 0.244, 0.722, and 0.606, respectively). Concerning the postoperative course, the day until first flatus, start of solid diet, time to drain removal, and the incidence of percutaneous drainage were similar between the stapler closure group and the PJ group.

U/L were similar between the stapler closure and PJ groups (P = 0.244, 0.722, and 0.606, respectively). Concerning the postoperative course, the day until first flatus, start of solid diet, time to drain removal, and the incidence of percutaneous drainage were similar between the stapler closure group and the PJ group. Association Between Clinically Relevant Pancreatic Fistula Based on the Thickness of the Pancreas Next, the association between the thickness of pancreas and clinically relevant pancreatic fistula was analyzed. In the stapler closure group, the thickness of the pancreas was similar in patients with or without clinically relevant pancreatic fistula (13.7 ± 3.9 mm vs 12.6 ± 5.0 mm, P = 0.516). In the PJ group, the thickness of the pancreas was also similar in patients with or without clinically relevant pancreatic fistula (13.8 ± 4.7 mm vs 11.4 ± 3.1 mm, P = 0.222). With a stratification based on 12-mm thickness at the resection site of the pancreas in the stapler closure group according to our previous reports,6,7 the incidence of clinically relevant pancreatic fistula associated with being below this cutoff point was not significantly different between the stapler closure group and the PJ group (P = 0.485). However, among patients with a pancreas thickness >12 mm, the incidence of clinically relevant pancreatic fistula occurred in 22.2% of the stapler closure group and 6.2% of the PJ group. PJ trended to decrease the incidence of clinical relevant pancreatic fistula after DP, although there was no significant difference (P = 0.080) (Table 4).

ts with a pancreas thickness >12 mm, the incidence of clinically relevant pancreatic fistula occurred in 22.2% of the stapler closure group and 6.2% of the PJ group. PJ trended to decrease the incidence of clinical relevant pancreatic fistula after DP, although there was no significant difference (P = 0.080) (Table 4). TABLE 4 Comparison of Clinical Relevant Pancreatic Fistula Between Staple Closure and PJ Based on the Thickness∗ of the Pancreas Clinically relevant pancreatic fistula† P Thickness of pancreas* ≦12 mm (n = 60) Stapler closure (n = 34) 4 (11.8%) 0.485 PJ (n = 26) 4 (15.4%) Thickness of pancreas* > 12 mm (n = 59) Stapler closure (n = 27) 6 (22.2%) 0.080 PJ (n = 32) 2 (6.2%) *Thickness at the resection site of pancreas. †Clinical pancreatic fistula is defined as pancreatic fistula grade B/C based on ISGPF.

Clinically relevant pancreatic fistula† P Thickness of pancreas* ≦12 mm (n = 60) Stapler closure (n = 34) 4 (11.8%) 0.485 PJ (n = 26) 4 (15.4%) Thickness of pancreas* > 12 mm (n = 59) Stapler closure (n = 27) 6 (22.2%) 0.080 PJ (n = 32) 2 (6.2%) *Thickness at the resection site of pancreas. †Clinical pancreatic fistula is defined as pancreatic fistula grade B/C based on ISGPF. DISCUSSION This prospective randomized multicenter study concluded that PJ of the pancreatic stump did not significantly reduce pancreatic fistula after DP compared with stapler closure. Three previous nonrandomized studies reported that PJ of the pancreatic stump was not associated with pancreatic fistula in any patients undergoing DP.8–10 The results of our study conflict with available results from previous studies in which PJ of the pancreatic stump was performed. There were 2 different points between our study and previous studies, namely, the technical approach in anastomosis of PJ and the use of octreotide. With regard to the technical approach, PJ in this study was performed in duct-to-mucosa fashion, using a single layer suture of resorbable interrupted stitches, while PJ in the previous studies8–10 was performed in a capsule-to-seromuscular fashion, using single-layer sutures after the main pancreatic duct was ligated. One RCT has reported that invagination PJ significantly reduced the incidence of pancreatic fistula after pancreaticoduodenectomy compared with duct-to-mucosa PJ,21 while 2 other RCTs have reported that duct-to-mucosa PJ and invagination PJ are comparable for the incidence of pancreatic fistula after pancreaticoduodenectomy.22,23 The appropriate procedure for PJ remains controversial. A second point of difference between our study and others was that prophylactic administration of octreotide was not performed in our study, although prophylactic octreotide was administrated on POD 5 to POD7 in previous studies.8–10 Somatostatin and its analogues have well-recognized inhibitory effects on pancreatic exocrine secretion. Two recent RCTs and 1 meta-analysis found that prophylactic somatostatin and its analogues significantly reduced the incidence of pancreatic fistula after pancreaticoduodenectomy.24–26 On the other hand, 4 other recent RCTs found that somatostatin analogues did not reduce pancreatic fistula after pancreaticoduodenectomy.27–30 No clear evidence exists in a favor of prophylactic administration of somatostatin and its analogues to reduce the incidence of pancreatic fistula.

pancreaticoduodenectomy.24–26 On the other hand, 4 other recent RCTs found that somatostatin analogues did not reduce pancreatic fistula after pancreaticoduodenectomy.27–30 No clear evidence exists in a favor of prophylactic administration of somatostatin and its analogues to reduce the incidence of pancreatic fistula. Therefore, we believe that duct-to-mucosa PJ and no administration of octreotide did not adversely affect the results of this study compared with the previous studies. However, most recent RCT published in 2014 concerning to somatostatin analogues demonstrated that prophylactic administration of pasireotide, a somatostatin analogue with a longer half-life, significantly reduced the incidence of clinically relevant pancreatic fistula.31 Pasireotide may change the treatment of pancreatic fistula in the near future. The most important reason that this study used duct-to-mucosa PJ was to decompress the main pancreatic duct and reduce pancreatic juice leakage from the branch pancreatic duct. One reasons for pancreatic fistula after DP is the increased resistance to the outflow of pancreatic juice toward the duodenum due to spasm of the sphincter of Oddi.32 Hashimoto and Traverso33 proposed that increased pancreatic ductal back pressure after DP was a risk factor for pancreatic stump leakage.

ncreatic duct. One reasons for pancreatic fistula after DP is the increased resistance to the outflow of pancreatic juice toward the duodenum due to spasm of the sphincter of Oddi.32 Hashimoto and Traverso33 proposed that increased pancreatic ductal back pressure after DP was a risk factor for pancreatic stump leakage. However, 1 RCT has reported that prophylactic transpapillary pancreatic stent for decompression of the pancreatic duct did not reduce pancreatic fistula after DP.34 In that RCT, pancreatitis due to insertion of pancreatic stent was not distinguished from postoperative infectious complications, so that the data including postoperative outcomes might lead to the negative results. In this RCT, we decided to perform the clinical trial intended the decompression of pancreatic duct without using the stent. Therefore, we hypothesized that decompression of pancreatic ductal back pressure by duct-to-mucosa PJ might prevent leakage of pancreatic juice from pancreatic stump.

ead to the negative results. In this RCT, we decided to perform the clinical trial intended the decompression of pancreatic duct without using the stent. Therefore, we hypothesized that decompression of pancreatic ductal back pressure by duct-to-mucosa PJ might prevent leakage of pancreatic juice from pancreatic stump. Aside from decompression of the main pancreatic duct, PJ was hypothesized to be superior to stapler closure for pancreatic stump because it was available regardless of the thickness of the pancreas resection site. A seromuscular-parenchymal anastomosis in PJ has a role in reducing pancreatic fistula because of the close proximity to the pancreatic stump despite the thickness of the pancreas resection site. Stapler closure of the pancreatic stump is technically easy and has become popular with many surgeons. However, stapler closure of pancreatic stump may not be useable for a thick pancreas.6,7 In this study, the rate of clinically relevant pancreatic fistula in the PJ group tended to be lower than that in the stapler closure group for patients with a thicker pancreas (ie, >12 mm; P = 0.08), but there was no statistically significance. This finding can be explained by a thick pancreatic parenchyma being easily torn by the compression of a stapler closure; the pancreas is a fragile organ. Therefore, using stapler closure for a thick pancreatic parenchyma may allow persistent extravasation of pancreatic juice by staple malformation. Hamilton et al35 demonstrated that resection with a stapler using mesh reinforcement greatly reduced pancreatic fistula. Also, Oláh et al18 showed the impact of covering the stapled pancreatic stump with seromuscular patch in significantly reducing pancreas-related complications. However, patients with thick or fibrotic pancreatic parenchyma were excluded from these RCTs. Therefore, PJ of pancreatic stump might offer a reduced risk of pancreatic fistula compared with stapler closure in a thick pancreas.

atic stump with seromuscular patch in significantly reducing pancreas-related complications. However, patients with thick or fibrotic pancreatic parenchyma were excluded from these RCTs. Therefore, PJ of pancreatic stump might offer a reduced risk of pancreatic fistula compared with stapler closure in a thick pancreas. Overall pancreatic fistula in the stapler group occurred in 37.7%. The incidence of overall pancreatic fistula in the stapler group was estimated to be 25%. The result in the stapler group was higher than estimated rate. However, the incidence of overall pancreatic fistula in DISPACT trial was 32%, which was approximately equal to this result. In addition, It remains still controversial which stapler cartridge is appropriate for stapler closure of the pancreatic stump. One study reported that stump closure using a white cartridge (compressible thickness to 1.0 mm) significantly decreased the incidence of clinically relevant pancreatic fistulas in comparison to that performed with a green cartridge (compressible thickness to 2.0 mm) (5% vs 31%),36 which might indicate that stump closure using a white cartridge may be useful to achieve a tighter seal of pancreatic stump in a thin pancreas. In contrast, a white cartridge may be traumatic to hard or thick pancreatic parenchyma. Thus, there has been no data to confirm which size of cartridge should be used for stump closure of distal pancreatectomy. A prospective randomized trial would be required to confirm which size of cartridge has a superior stump closure.

ontrast, a white cartridge may be traumatic to hard or thick pancreatic parenchyma. Thus, there has been no data to confirm which size of cartridge should be used for stump closure of distal pancreatectomy. A prospective randomized trial would be required to confirm which size of cartridge has a superior stump closure. PJ of the pancreatic stump may result in more dangerous complications than staple closure due to intestinal juice. Moreover, additional procedures involving the intestine may cause extra complications related to intestinal resection. However, the incidence of intra-abdominal abscess requiring percutaneous drainage was 8.1% (5 of 62) in the PJ group compared with 4.9% (3 of 61) in the stapler closure group. PJ resulted in no significant increase of intra-abdominal abscess compared with stapler closure (P = 0.331). In addition, the incidence of ileus and postoperative course, including the day until first flatus and start of solid diet, were comparable between stapler closure and PJ. In conclusion, this study could not evaluate the superiority of PJ of pancreatic stump during DP to reduce pancreatic fistula compared with staple closure. However, PJ for pancreatic stump might offer a potential reduction of pancreatic fistula in cases with a thick pancreas. RCT by stratification of a thick pancreas is required to confirm the impact of PJ of pancreatic stump to reduce pancreatic fistula. Disclosure: The authors declare no conflict of interest.

Iron deficiency (ID) and iron deficiency anemia (IDA) are common conditions affecting a quarter of the world's population.1 Aside from age, socioeconomic circumstances, poor nutrition, and pregnancy, many pathological states frequently lead to iron depletion. In patients presenting for noncardiac surgery, ID with or without anemia is found in up to 39% of patients.2 In certain subgroups, like colorectal cancer or patients with heavy menstrual bleeding, the occurrence of preoperative anemia has been found to be as high as 57 %.3,4 The cause of the ID is either disease related, for which the patient is being treated, or therapy related, as for many individuals on nonsteroidal anti-inflammatory drugs.5 Anemia, ABT, and perioperative significant blood loss have all been established as adversely impacting clinical outcomes.6 Perioperatively, anemic patients frequently are exposed to allogeneic transfusion events, increased rates of infections, increased numbers of cardiac complications, an increased number of days in hospital, and more deaths.2,7–9 Anemia is often corrected by ABT, despite the evidence that even small amounts of transfused allogeneic red cells have a significant negative effect on morbidity, mortality,10 and reduces cancer-related survival and overall survival in colorectal cancer patients.9 Transfusion-associated hazards and the noninferiority of restrictive transfusion approaches have been demonstrated, challenging traditional transfusion practice.11–13

ls have a significant negative effect on morbidity, mortality,10 and reduces cancer-related survival and overall survival in colorectal cancer patients.9 Transfusion-associated hazards and the noninferiority of restrictive transfusion approaches have been demonstrated, challenging traditional transfusion practice.11–13 Comprehensive patient blood management (PBM) programs offering effective approaches for minimizing perioperative blood loss and optimized patient care have been designed and implemented in some countries.14,15 However, due to evidence gaps, translational delays for existing evidence, and ongoing skepticism, transfusion practices continue to vary considerably among clinicians.16 Preoperative optimization of anemia seems to be a key aspect of PBM.6,16,17 In particular, patients scheduled for major surgery and with medical conditions often associated with ID should be assessed at least 4 weeks before surgery to allow clinicians to interpret blood results with a window of opportunity to act and correct reversible hemopoietic deficiencies.17–19

a key aspect of PBM.6,16,17 In particular, patients scheduled for major surgery and with medical conditions often associated with ID should be assessed at least 4 weeks before surgery to allow clinicians to interpret blood results with a window of opportunity to act and correct reversible hemopoietic deficiencies.17–19 Both oral and IV iron have been shown to correct ID and IDA,20,21 but neither has become standard practice,22 nor has the ideal timing of preoperative intervention for IV iron been determined.20,23 Oral iron replacement is in many instances poorly tolerated, ineffective, or even detrimental.24,25,26 There is increasing evidence that in select patient groups presenting for elective surgery and in urgent cases, treatment with IV iron might benefit the patient and should result in a reduction of RBC transfusion and transfusion-related adverse events.17,27 Should an increased use of IV iron for at-risk patients have the predicted effect of improved hematological parameters and restored iron levels, the impact would translate to significant benefits for the individual in the immediate postoperative period and the weeks after hospital discharge. METHODS This was a randomized controlled trial. The protocol was approved by the study hospital's human research ethics committee and registered with the Australian New Zealand Clinical Trials Registry (ACTRN12611000387921).

Both oral and IV iron have been shown to correct ID and IDA,20,21 but neither has become standard practice,22 nor has the ideal timing of preoperative intervention for IV iron been determined.20,23 Oral iron replacement is in many instances poorly tolerated, ineffective, or even detrimental.24,25,26 There is increasing evidence that in select patient groups presenting for elective surgery and in urgent cases, treatment with IV iron might benefit the patient and should result in a reduction of RBC transfusion and transfusion-related adverse events.17,27 Should an increased use of IV iron for at-risk patients have the predicted effect of improved hematological parameters and restored iron levels, the impact would translate to significant benefits for the individual in the immediate postoperative period and the weeks after hospital discharge. METHODS This was a randomized controlled trial. The protocol was approved by the study hospital's human research ethics committee and registered with the Australian New Zealand Clinical Trials Registry (ACTRN12611000387921). We randomly allocated participants (1:1) to either perioperative intravenous (IV) iron administration (intervention) or usual care. Randomization followed a computer-generated number sequence and allocation was conducted by telephone. The surgeon performing the operation was informed of patient participation in the study but group allocation was not revealed.

to either perioperative intravenous (IV) iron administration (intervention) or usual care. Randomization followed a computer-generated number sequence and allocation was conducted by telephone. The surgeon performing the operation was informed of patient participation in the study but group allocation was not revealed. We screened 626 patients scheduled for abdominal surgery for the presence of IDA between August 2011 and November 2014. After informed written consent, patients eligible for inclusion (>18 yrs with IDA, ferritin <300 mcg/L, transferrin saturation <25%, Hb <12.0 g/dL for women, Hb <13.0 g/dL for men) were randomized between 4 and 21 days before surgery into 2 groups. Owing to this wide range in the preoperative period between patients, a standard approach was used to assess transfusion events in the preoperative period, including any transfusion administered in the 21 days before surgery. Patients in the intervention group received IV ferric carboxymaltose, given as a single dose over 15 minutes, before surgery (simplified dosing protocol; 15 mg/kg bodyweight to a maximum dose of 1000 mg). Postoperatively, within 2 days of surgery, intervention group participants received 0.5 mg of ferric carboxymaltose per recorded 1 mL of blood loss, if blood loss was at least 100 mL. Blood loss was measured as accurately as possible by recording suction bottle volume and weighing packs at the end of the operation. Patients in the usual care group received perioperative care, including anemia management, provided by the primary care physician or surgical home team. Usual care provided included no treatment, continued observations, oral iron recommendations, and ABT. At the time of initiation of the study, IV iron was not considered usual care; however, prescription and administration was not disallowed.

mia management, provided by the primary care physician or surgical home team. Usual care provided included no treatment, continued observations, oral iron recommendations, and ABT. At the time of initiation of the study, IV iron was not considered usual care; however, prescription and administration was not disallowed. In the institution, the prescription and administration of the intervention was facilitated by the anesthetic team. Baseline testing of the Short Form Health Survey (SF36) was conducted at study entry.28 Follow-up of participants was scheduled for 4 weeks after surgery. The SF36 and screening bloods were repeated at this time. Patients found to have noteworthy ID or IDA at follow-up, irrespective of group allocation, were referred to their General Practitioner for ongoing management. Quality Control Procedure A trial information session was given to the departmental members involved before commencing the study. Regular refreshers were scheduled to assure protocol knowledge and adherence. The multidisciplinary composition of the research team facilitated this process. Follow-up and data entry were meticulously conducted by a research assistant, and primary care physician follow-up and care initiated when necessary to assure patient safety.

shers were scheduled to assure protocol knowledge and adherence. The multidisciplinary composition of the research team facilitated this process. Follow-up and data entry were meticulously conducted by a research assistant, and primary care physician follow-up and care initiated when necessary to assure patient safety. Outcomes The primary endpoint was incidence of ABT. Secondary endpoints included hemoglobin (Hb) on admission, Hb difference from randomization to admission, ICU admission, perioperative morbidity (defined as new onset infection, respiratory failure, renal impairment, deep venous thrombosis), discharge Hb, length of stay, Hb at follow-up, Hb difference from discharge to follow-up, iron status, 30-day mortality, and quality of life (QoL). QoL score was scaled from 36 to 160, with lower scores reflecting poorer well-being.

as new onset infection, respiratory failure, renal impairment, deep venous thrombosis), discharge Hb, length of stay, Hb at follow-up, Hb difference from discharge to follow-up, iron status, 30-day mortality, and quality of life (QoL). QoL score was scaled from 36 to 160, with lower scores reflecting poorer well-being. Statistical Analysis The sample size of this study was calculated for the primary outcome parameter (perioperative allogeneic transfusion event). To reduce the risk of a perioperative allogeneic transfusion event from 30% to 15% (a 50% risk reduction) with a power of β = 0.8 and a significance level of α = 0.05, it was determined that a total of 121 patients in each group would be needed. To account for possible dropouts, we intended to include 134 patients per group. The power calculation was performed using nQuery Advisor Version 7.0 (Statistical Solutions, Saugus, MA). Parametric data were tested with one-way ANOVA, and are presented as mean and standard error of the mean or as mean and 95% confidence intervals. Nonparametric data were tested with Mann-Whitney U tests, and are reported as either median (IQR) or median (minimum–maximum), as indicated. Categorical data were analyzed with the 2-tailed Pearson χ2 test, and are presented accordingly as number and percent of total. Statistical analysis was performed using SPSS software version 17.0 (SPSS Inc, Chicago, IL).

Whitney U tests, and are reported as either median (IQR) or median (minimum–maximum), as indicated. Categorical data were analyzed with the 2-tailed Pearson χ2 test, and are presented accordingly as number and percent of total. Statistical analysis was performed using SPSS software version 17.0 (SPSS Inc, Chicago, IL). RESULTS An early interim data analysis was requested following concerns raised by the clinical investigator team after high rates of RBC transfusion, considered to be an independent risk factor for adverse clinical outcomes, noted after the 4-week follow-up in a subset of patients. This was performed by an independent statistician on the interim data-monitoring committee with the data blinded (intervention group n = 32, usual care group n = 26). The results of the interim analysis were forwarded to 2 independent experts in the field to assess safety concerns. Enrolment continued while waiting for a response. There was disagreement among the assessors, and a third independent expert opinion was sought. Based on advice from 2 of the 3 independent experts, the study was terminated early due to higher than expected rates of poor outcome in the usual care group.

safety concerns. Enrolment continued while waiting for a response. There was disagreement among the assessors, and a third independent expert opinion was sought. Based on advice from 2 of the 3 independent experts, the study was terminated early due to higher than expected rates of poor outcome in the usual care group. At the time of study termination, 72 eligible patients were enrolled and randomized (intervention group n = 40, usual care group n = 32) (see Supplemental Digital Content eFigure 1). Group characteristics are shown in Table 1, and the type of surgery for the patient groups is shown in Supplementary Table 1. Cancer was the underlying condition in 73% of group 1 patients and 85% of patients in the usual care group. The overall transfusion rate in the study was 20.8%. Ten patients in the usual care group (10/32 = 31.25%) were transfused vs 5 in the intervention group (5/40 = 12.5%), equating to a 60% relative reduction in transfusions between the 2 groups shown in Table 2. There were no intraoperative RBC transfusions in the intervention group compared with 5 in the usual care group (P = 0.014) and a significant reduction in the number of total perioperative ABT events in group 1 (5/40, 12.5 %) compared with group 2 (17/32, 53 %), P < 0.0003. The median number of units per transfused patient was also decreased in the intervention group (2 compared with 3 in the usual care group; P = 0.016; Table 2). There was no difference in the rationale for transfusion between the 2 groups with the majority being performed due to low hemoglobin (Supplementary Table 2). The median IV iron dose administered to participants in the intervention group was 1200 mg (IQR 1088–1363). Five participants in the usual care group were given a median IV iron dose of 1800 mg (IQR 1467–2000). Any participant receiving IV iron had a maximum of 2 infusions. No serious adverse event resulted from the iron infusion. Three participants suffered the following mild adverse events: headache, light-headedness, and back pain. The latter settled with simple analgesics. Hb levels across study period and other important secondary outcome parameters are shown in Tables 3 and 4. Hb values were not different at randomization and improved by 0.8 g/dL in group 1 and by 0.1 g/dL in the usual care group (P = 0.01) by the day of admission.

back pain. The latter settled with simple analgesics. Hb levels across study period and other important secondary outcome parameters are shown in Tables 3 and 4. Hb values were not different at randomization and improved by 0.8 g/dL in group 1 and by 0.1 g/dL in the usual care group (P = 0.01) by the day of admission. Despite higher transfusion rate in the usual care group, there were no differences between groups in discharge Hb (10.3 vs 10.2 g/dL for the intervention group and usual care group, respectively). However, Hb increased by 1.9 g/dL in the intervention group and 0.9 g/dL in the usual care group (P = 0.01) from the time of discharge to follow-up and was significantly higher at 4 weeks postsurgery (12.2 g/dL compared with 11.1 g/dL in the usual care group, P < 0.001). Length of stay was shortened by 3 days in the intervention group compared with the usual care group (6 vs 9 d, P = 0.05). There was no significant difference in morbidity or mortality (Table 4). QoL scores were higher at baseline for the intervention group; however, score reduction was equal between the groups. TABLE 1 Group Characteristics Intervention n = 40 Usual Care n = 32 Demographics Age 64 ± 15 68 ± 15 Height, cm 166 ± 10 167 ± 10 Weight, kg 86 ± 27 88 ± 20 BMI, kg/m2 30 ± 8 31 ± 7 ASA 2 17 (42.5) 17 (53.1) ASA 3 22 (55) 15 (46.9) ASA 4 1 (2.5) 0 Sex (male/female) 19/21 17/15 Surgery Days before surgery* 8 (6–13) 10 (5–15) Estimated blood loss during surgery* 360 (200–700) 300 (100–512) *Data are median (IQR).

8 ± 15 Height, cm 166 ± 10 167 ± 10 Weight, kg 86 ± 27 88 ± 20 BMI, kg/m2 30 ± 8 31 ± 7 ASA 2 17 (42.5) 17 (53.1) ASA 3 22 (55) 15 (46.9) ASA 4 1 (2.5) 0 Sex (male/female) 19/21 17/15 Surgery Days before surgery* 8 (6–13) 10 (5–15) Estimated blood loss during surgery* 360 (200–700) 300 (100–512) *Data are median (IQR). BMI indicates body mass index; ASA, physical status classification system; IQR, interquartile range. TABLE 2 Primary Outcome; Perioperative Red Blood Cell Transfusions Transfusion Events Occurring in Each Period Intervention n = 40 Usual Care n = 32 P Preoperative 0 2 (6%) 0.190 Units transfused 0 7 Intraoperative 0 5 (16%) 0.014 Units transfused n.a. 7 Postoperative 5 (12%) 10 (31%) 0.079 Units transfused 8 18 Total number of Patients transfused 5 (12.5%) 10 (31.25%) 0.079 Units transfused 8 32 Total number of transfusion events 5 17 <0.001 Total units/patient (median, minimum–maximum) 0 (0–2) 0 (0–5) 0.021 Total units/transfused patient (median, minimum–maximum) 2 (1–2) 3 (2–5) 0.016 Transfusion events presented in the table are differentiated into time of transfusion and total events experienced in the perioperative period. The preoperative period was defined as the 3 weeks before surgery; data are n (%) unless otherwise stated. TABLE 3 Secondary Outcomes; Hematological Indices Across Study Period

Transfusion Events Occurring in Each Period Intervention n = 40 Usual Care n = 32 P Preoperative 0 2 (6%) 0.190 Units transfused 0 7 Intraoperative 0 5 (16%) 0.014 Units transfused n.a. 7 Postoperative 5 (12%) 10 (31%) 0.079 Units transfused 8 18 Total number of Patients transfused 5 (12.5%) 10 (31.25%) 0.079 Units transfused 8 32 Total number of transfusion events 5 17 <0.001 Total units/patient (median, minimum–maximum) 0 (0–2) 0 (0–5) 0.021 Total units/transfused patient (median, minimum–maximum) 2 (1–2) 3 (2–5) 0.016 Transfusion events presented in the table are differentiated into time of transfusion and total events experienced in the perioperative period. The preoperative period was defined as the 3 weeks before surgery; data are n (%) unless otherwise stated. TABLE 3 Secondary Outcomes; Hematological Indices Across Study Period Intervention n = 40 Usual Care n = 32 P Hemoglobin values, g/dL* Randomization 10.7 ± 1.3, n = 40 10.6 ± 1.4, n = 32 0.76 Admission 11.5 ± 1.3, n = 36 10.7 ± 1.7, n = 29 0.12 Difference between randomization and admission 0.8 ± 0.8, n = 36 0.1 ± 1.3, n = 29 0.01 Discharge 10.3 ± 1.3, n = 37 10.2 ± 0.9, n = 31 0.31 4 wk 12.2 ± 1.2, n = 36 11.1 ± 1.2, n = 28 <0.001 Postdischarge change (4 wk minus discharge value) 1.9 ± 1.4, n = 36 0.9 ± 1.4, n = 28 0.01 Iron status† Ferritin at randomization, μg/L 19 (6–48), n = 40 37 (11–82), n = 32 0.06 Ferritin at 4 wk, μg/L 248 (137–546), n = 36 99 (35–228), n = 27 0.002 Transferrin saturation at randomization, % 6 (3–10), n = 40 9 (7–15), n = 32 0.03 Transferrin saturation at 4 wk, % 21 (16–26), n = 36 14 (7–18), n = 27 0.003 CRP† CRP at randomization, mg/L 7.2 (2.9–19.3), n = 40 7.7 (2.6–16.8), n = 32 0.99 CRP at 4 wk, mg/L 5.8 (2.3–12.6), n = 36 11 (3.1–23), n = 27 0.18 Sample size varies due to missing values at each time point.

10), n = 40 9 (7–15), n = 32 0.03 Transferrin saturation at 4 wk, % 21 (16–26), n = 36 14 (7–18), n = 27 0.003 CRP† CRP at randomization, mg/L 7.2 (2.9–19.3), n = 40 7.7 (2.6–16.8), n = 32 0.99 CRP at 4 wk, mg/L 5.8 (2.3–12.6), n = 36 11 (3.1–23), n = 27 0.18 Sample size varies due to missing values at each time point. *Data presented as mean ± SD. †Data are expressed as median (IQR). TABLE 4 Other Secondary Outcomes of Interest Intervention n = 40 Usual Care n = 32 P Length of stay, d* 6 (1–19) 9 (1–23) 0.05 Infection† 4 (10%) 5 (16%) 0.5 Respiratory failure† 3 (7.5%) 3 (9%) 0.99 Renal impairment† 1 (2.5%) 1 (3%) 0.99 DVT† 0 1 (3%) 0.45 Readmission† 6 (15%) 3 (9%) 0.72 Discharged on oral iron† 5 (12.5%) 1 (3%) 0.22 Death† 1 (2.5%) 0 0.99 QoL (presurgery/intervention)‡ 104 ± 15 96 ± 18 0.02 QoL (4 wk postsurgery)‡ 96 ± 14 90 ± 26 0.24 Difference in QoL (pre–post)‡ 8 ± 18 6 ± 17 0.70 QoL score scale; 36–160, lower scores reflecting poorer well-being. *Median (minimum–maximum). †Data are n (%). ‡Data are expressed as mean ± SD unless otherwise stated.

Intervention n = 40 Usual Care n = 32 P Length of stay, d* 6 (1–19) 9 (1–23) 0.05 Infection† 4 (10%) 5 (16%) 0.5 Respiratory failure† 3 (7.5%) 3 (9%) 0.99 Renal impairment† 1 (2.5%) 1 (3%) 0.99 DVT† 0 1 (3%) 0.45 Readmission† 6 (15%) 3 (9%) 0.72 Discharged on oral iron† 5 (12.5%) 1 (3%) 0.22 Death† 1 (2.5%) 0 0.99 QoL (presurgery/intervention)‡ 104 ± 15 96 ± 18 0.02 QoL (4 wk postsurgery)‡ 96 ± 14 90 ± 26 0.24 Difference in QoL (pre–post)‡ 8 ± 18 6 ± 17 0.70 QoL score scale; 36–160, lower scores reflecting poorer well-being. *Median (minimum–maximum). †Data are n (%). ‡Data are expressed as mean ± SD unless otherwise stated. DISCUSSION This first RCT on managing preoperative anemia in abdominal surgery, involving only patients with confirmed IDA, demonstrates the important role for IV iron in perioperative PBM. The results also highlight the ongoing mismanagement of a treatable condition despite the well-known negative impact of IDA.29,30 In addition, it also confirms the ongoing overuse of ABT as a default treatment approach31–33 regardless of the well-described transfusion-related risks34 and the safety of restrictive transfusion practices.10,12,13 We also report that although Hb levels were equivalent in the 2 groups at discharge, they were 1 g/dL higher in the treatment group compared with the usual care group at 4 weeks after surgery. This demonstrates that perioperative iron repletion has substantial benefit in the postoperative recovery period, potentially due to the iron repletion allowing bone marrow to increase erythropoiesis, compared with transfused RBCs which are rapidly cleared from the circulation and have a shorter lifespan than normal RBCs.35 The superiority of IV iron over oral or no iron in reducing ABT was previously demonstrated in other clinical setting and extensively discussed in a recent review by Muñoz.36

ease erythropoiesis, compared with transfused RBCs which are rapidly cleared from the circulation and have a shorter lifespan than normal RBCs.35 The superiority of IV iron over oral or no iron in reducing ABT was previously demonstrated in other clinical setting and extensively discussed in a recent review by Muñoz.36 Transfusion triggers and the appropriateness of ABT administration were the focus of perioperative transfusion management at the time when we designed this study. Our aim was to determine whether perioperative IV iron, administered within 4 to 21 days before substantial abdominal surgery, would lead to a significant reduction in transfusion events. We anticipated that we would demonstrate that this intervention would not only obviate ABT, but also correct underlying iron deficits, facilitating better recovery and outcomes. Since the commencement of our study, the importance of correcting preoperative ID has been more widely accepted as an appropriate standard of care, strengthening our hypothesis.6,17,18,21,37 The value of preoperative correction of IDA has thus become a cornerstone of PBM guidelines around the world.14,15,38

d outcomes. Since the commencement of our study, the importance of correcting preoperative ID has been more widely accepted as an appropriate standard of care, strengthening our hypothesis.6,17,18,21,37 The value of preoperative correction of IDA has thus become a cornerstone of PBM guidelines around the world.14,15,38 However, data monitoring of our participants indicated that a large proportion of enrolled subjects in the usual care group were transfused with RBC to correct anemia but received no treatment for their ID. RBC transfusion is considered to be an independent risk factor for adverse clinical outcomes.10,31,39 Recognition of this situation and the ethical responsibility to our participants prompted an interim analysis, and the seeking of advice from impartial experts to assess whether early termination of the study was recommended scientifically and ethically. Enrolment in the study continued during the assessment and decision-making process. After definitive analysis of the expert opinion, it was deemed that the study should be terminated in the interest of the patients. As early as 1985, influenced by the AIDS epidemic, strategies for avoiding or minimizing ABT were published.40 Compelling evidence on the importance of anemia and blood management from the last 15 years 13,37,41 put PBM on the agenda and illustrated how PBM should be carried out. Logically, one would expect that “standard care” had moved on. However, our results show that the translational gap is huge and that anemia management has some way to go in clinical practice.19,33,34,42–45

od management from the last 15 years 13,37,41 put PBM on the agenda and illustrated how PBM should be carried out. Logically, one would expect that “standard care” had moved on. However, our results show that the translational gap is huge and that anemia management has some way to go in clinical practice.19,33,34,42–45 The assessment of adequate iron stores can be difficult. With ferritin levels influenced by chronic disease and/or inflammation, ID may be masked. Therefore, screened subjects were included with ferritin levels of less than 300 mcg/L in our study, as recommended in a consensus statement on the role of IV iron in perioperative anemia management.46 The distribution of ferritin levels was essentially the same in the 2 groups of participants; 48% (intervention) and 40% (usual care) presented with profound ID, demonstrated by ferritin levels less than 30 mcg/L. Despite sometimes longstanding and previously diagnosed IDA, only 3 patients in our entire cohort had been prescribed oral iron replacement therapy within the 6 weeks before surgery. Only 1 patient in the usual care group was treated with IV iron pre- and postoperatively, and 4 received IV iron while in hospital. IV iron was not considered usual care, at the time of study commencement, nonetheless was not prohibited. Patients were randomized in our study between 8 and 10 days before admission. Although it is desirable for ID to be corrected in a timely manner, the study establishes that a successful “rescue” intervention is available and effective at a later stage, even for those with profound IDA. Our results support a proposed “opportunity” approach, discussed in a recent review article by Muñoz et al, and suggested earlier based on results from pooled data by the same author.17,47

establishes that a successful “rescue” intervention is available and effective at a later stage, even for those with profound IDA. Our results support a proposed “opportunity” approach, discussed in a recent review article by Muñoz et al, and suggested earlier based on results from pooled data by the same author.17,47 In addition to risk minimization and outcome improvement, our findings might have significant economic implications. According to the Australian Institute of Health and Welfare, 15,840 patients were diagnosed with bowel cancer in Australia in 2012 and many had to undergo abdominal surgery. Cancer patients made up the majority of our cohort. The patients randomized to receive pre- and postoperative IV iron left hospital 3 days earlier. We suggest that this earlier discharge was due to treatment of ID with IV iron, thus minimizing the associated risks of this exposure. Although a cost analysis was beyond the scope of this research, we propose that this would result in a significant cost savings, offsetting the initial expenditure of screening.

arlier. We suggest that this earlier discharge was due to treatment of ID with IV iron, thus minimizing the associated risks of this exposure. Although a cost analysis was beyond the scope of this research, we propose that this would result in a significant cost savings, offsetting the initial expenditure of screening. Limitations Early termination is the main limitation of our study. However, ethical concerns were paramount, and we made the necessary decision in the interest of our patients. Although more cases would strengthen the statistics, it was not anticipated that the conclusions would change. In our view it would have been unethical to have iron-deficient patients in a control group at increased risk of receiving a blood transfusion. A serious hazard from an ABT resulting in morbidity or mortality in a control patient would be difficult to defend. Another limitation is that we performed simple randomization instead of block randomization. This was apparent at the time of the interim analysis and the final analysis after stopping the study. Block randomization would have achieved a more equal balance in the allocation of participants. In this study, 5 participants randomized to the usual care group received IV iron as part of their standard care. Although this may have influenced the results, the final analysis between groups would then represent a more conservative analysis of the effects of IV iron. This change in standard care of iron deficient patients further adds clarity to the decision for early termination.

ceived IV iron as part of their standard care. Although this may have influenced the results, the final analysis between groups would then represent a more conservative analysis of the effects of IV iron. This change in standard care of iron deficient patients further adds clarity to the decision for early termination. CONCLUSIONS In conclusion, the administration of IV iron in the perioperative setting resulted in a significant reduction of RBC transfusion, significant Hb improvement from the time of randomization to admission, shorter hospital stays, and enhanced restoration of iron stores and Hb at 4 weeks after surgery. Usual care failed the majority of participating patients, leaving them untreated with a treatable condition. Supplementary Material Supplemental Digital Content Acknowledgments The authors thank the staff working in the Day Procedure Unit for accommodating the iron infusions, and Dr David Huang for critically reviewing the manuscript. Funding: University of Adelaide; Strategic Initiative Fund. Contributions: BF, IW, PP, and EMM did the study and collected data. Conception and design: BF, RS, NAH, and EMM. NAH and BF did the primary analysis, and all authors contributed to the interpretation of data and preparation of the manuscript. Final approval: All authors. The author reports no conflicts of interest. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Website (www.annalsofsurgery.com).

d to specific subtypes of PG and PJ. Only 2 RCTs with contradictory results report on postoperative pancreatic function measured during follow-up of 3 to 12 months: the Egyptian trial26 reports worse and the Spanish trial25 reports better pancreatic function. None of the RCTs report on quality of life during follow-up. Here we present data collected at 14 high-volume centers for pancreatic surgery in Germany from the currently largest multicenter randomized trial comparing PG with PJ with respect to perioperative complications and long-term pancreatic function and quality of life. PATIENTS AND METHODS Study Design, Hypothesis, and Inclusion Criteria The RECOnstruction after PANCreatoduodenectomy Study (RECOPANC) was designed as a randomized, controlled, observer- and patient-blinded multicenter trial with 2 parallel treatment arms (PG and PJ) (see Supplemental Digital Content, available at). The hypothesis was that the rate of clinically relevant POPF is lower after PG. Inclusion criteria were planned pancreatoduodenectomy at one of the participating academic centers and age more than 18 years. Exclusion criteria were participation in interfering clinical trials and expected lack of compliance. With the rationale to increase willingness of participating surgeons to recruit patients and to achieve greater generalizability of the results, we did not restrict PG or PJ to a special technique. Fourteen German academic centers (RECOPANC Trial Group27) with a median case load of 78 major pancreatic resections per year (range: 29–499, figures for year 2012 from the Association of German University Clinics, http://www.uniklinika.de) participated in the trial.

we did not restrict PG or PJ to a special technique. Fourteen German academic centers (RECOPANC Trial Group27) with a median case load of 78 major pancreatic resections per year (range: 29–499, figures for year 2012 from the Association of German University Clinics, http://www.uniklinika.de) participated in the trial. Primary Endpoint and Sample Size POPF is defined by ISGPS as the occurrence of amylase activity in abdominal drain fluid of 3 times the upper serum limit on postoperative day 3 or later.28,29 In brief, grade A fistula is self-limited and does not need specific treatment, grade B requires medical or invasive interventional treatment, and grade C leads to reoperation and/or severe secondary complications. The primary endpoint chosen for this trial was clinically relevant POPF, that is, ISGPS grade B or C, with the modification that application of somatostatin analogues was not considered a criterion for grading. The primary endpoint was assessed on postoperative day 3 at hospital discharge and on postoperative day 30 to detect all POPFs. Based on the prior assumption of a POPF B/C rate of 6% and 16% with PG and PJ, respectively, α = 5% and β = 20%, a sample size of 153 per treatment arm (PG vs PJ) was calculated with the 2-sided χ2 test. An adaptive interim analysis of the primary endpoint according to Bauer and Koehne30 was planned after recruitment of 152 patients to allow for premature trial termination (with 1-sided stopping boundaries of P < 0.0038) and sample size recalculation.

er treatment arm (PG vs PJ) was calculated with the 2-sided χ2 test. An adaptive interim analysis of the primary endpoint according to Bauer and Koehne30 was planned after recruitment of 152 patients to allow for premature trial termination (with 1-sided stopping boundaries of P < 0.0038) and sample size recalculation. Secondary Endpoints and Follow-up Secondary surgical endpoints were death, relaparotomy, completion pancreatectomy, anastomotic leak other than pancreatic fistula, wound infection, delayed gastric emptying, postpancreatectomy hemorrhage according to the ISGPS definitions,31,32 intra-abdominal abscess requiring invasive treatment, operation time (skin incision to skin closure), and postoperative hospital stay. Further secondary endpoints included septic shock, respiratory failure, deep venous thrombosis, lung embolism, myocardial infarction, and stroke. Pancreatic endocrine and exocrine functions and quality of life were evaluated in long-term follow-up at baseline, 6 and 12 months after the operation by the validated European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC-QLQ) C30 and the pancreatic cancer module PAN26.33,34

e. Pancreatic endocrine and exocrine functions and quality of life were evaluated in long-term follow-up at baseline, 6 and 12 months after the operation by the validated European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC-QLQ) C30 and the pancreatic cancer module PAN26.33,34 Randomization and Blinding Center-based block randomization was performed by the participating centers using a centralized Web-based tool (Randomizer Software, Institute for Medical Informatics, Statistics and Documentation of the Medical University of Graz, www.randomizer.at) with allocation concealment. To avoid a possible intraoperative selection of low-risk patients,23 randomization was performed preoperatively. Obviously, the surgeons were not blinded concerning the intervention. Therefore, blinded observers at the participating centers assessed the primary endpoint. Patients were kept blinded regarding the intervention and unblinded only in the case of emergencies where necessary.

randomization was performed preoperatively. Obviously, the surgeons were not blinded concerning the intervention. Therefore, blinded observers at the participating centers assessed the primary endpoint. Patients were kept blinded regarding the intervention and unblinded only in the case of emergencies where necessary. Ethical Approval, Safety, and Registration The study protocol was approved by the local ethics committees of the participating centers and carried out according to the rules of Good Clinical Practice and the Declaration of Helsinki.35 Written informed consent was obtained from each patient. An independent institution served as the Data Safety Monitoring Board and was responsible for on-site clinical monitoring, source data verification, and management of severe adverse event reports (Center for Clinical Studies, Freiburg, Germany). The trial was assigned a Universal Trial Number (UTN U1111-1117-9588) and registered in the German Trials Register (DRKS 00000767) on March 23, 2011. The study protocol was published in Trials.27

, source data verification, and management of severe adverse event reports (Center for Clinical Studies, Freiburg, Germany). The trial was assigned a Universal Trial Number (UTN U1111-1117-9588) and registered in the German Trials Register (DRKS 00000767) on March 23, 2011. The study protocol was published in Trials.27 Statistical Analysis The primary endpoint was analyzed according to the intention-to-treat principle (see Supplemental Digital Content, available at). A multivariate logistic regression model adjusting a priori for age, center, surgeon volume/experience, and pancreatic texture was applied to compare POPF rates in both treatment groups. Missing values for the primary endpoint were replaced by imputed case analysis according to Higgins et al.36 Exploratory analysis was planned for secondary endpoints. SAS software 9.1 (SAS 9.1 software, SAS, Cary NC) and 2-sided tests were used for all calculations.